Cruces — Anselm of Laon and his school, Liber Exodus

These are the edition's working notes, published as they were written. They record, work by work, where Migne's plate is defective or the reading uncertain: the printed form, what the English does with it, and why. They are not a groomed apparatus. Where a note argues with itself, withdraws an earlier decision, or leaves a question open, that is the record of how the reading was actually reached, and it is left standing — a conjecture that is visible can be checked, and one that has been tidied away cannot.

The English never silently repairs the page. A construable but wrong reading is rendered literally and the conjecture is recorded here, never applied to the text. How the English is produced and checked →

Merged 2026-09-08 from nine stint files. The merge is a required step, not a tidy-up: nine agents translated this book concurrently and none could see another's choices, so the work of merging is reconciling convention drift, not concatenating. What follows is the work-wide record; each stint's own file is preserved below it verbatim.

§0 · WORK-WIDE CONVENTIONS — settled at merge, and binding on any later sweep

Do not churn these. Each was fixed for a reason recorded beside it; a later pass that "regularises" one will break the gloss that depends on it.

### The lemma

⭐ Exodus is an ITALIC book, and it is at the far end of the range. The whole work contains one guillemet — an opening « at 0192A with no » anywhere — where Genesis (8950), the same commentary on the neighbouring book, sets 1,207 lemmata in guillemets. That single mark was read at the plate (Gallica f96, corners 191/192): Migne opens a quotation of Jn 1:27 and never closes it, and our Latin has it exactly as he prints it. No [sic:], no patch, and it is not a lemma boundary anyone may "balance."
The lemma renders MIGNE'S wording, never a conventional English version. 123 [var:] mark the places where his lemma or proof-text departs from the Clementine, and each was plate-confirmed before firing.
The VERS. n.-- address is kept exactly as printed. ⚑ Small caps decide whose a defect is: Vers. 9. at 0202D is OURS and was patched; VERS 20. at 0201C, missing its period, is HIS and must never be patched.

### Vocabulary fixed for the whole work

⭐ praeceptum = "precept"; mandatum = "commandment." Ruled at merge and swept. The evidence is Migne's own page: Isidore uses both words four lines apart at 0254C, Origen's enumerated series is mandatum and Rabanus's is praeceptum. Under one English word for both, that paragraph reads as one word seven times and two authors' expositions merge into a single enumeration that keeps restarting. Swept at five sites — 0002 (×3, praeceptum regis Aegypti) and 0009 (×2, the Decalogue's own enumeration).
⭐ tabula does TWO jobs in this book and both renderings are correct. The Law's tabulae are "tables" (19 sites); the tabernacle's tabulae are "boards" (26 sites). The board/table split was already clean at 13/13 and 9/9 in the tabernacle chunks. cruces-0030 proposed "tablet" for the Law and that is overruled: 14 sites across six chunks already read "tables", it matches the Douay wording the [var:] notes themselves quote, and the sweep churns one chunk instead of six. 0039 was swept from "tablet" to "table(s)".
The tabernacle's technical vocabulary is fixed in cruces-0030.md §6 — 40 terms, each with its reason — and chunks 0035–0043 follow it. Where the gloss's own argument decided against the idiomatic English, the argument won: propitiatorium = propitiatory (0267C is an etymology from propitiari; "mercy seat" destroys it), panes propositionis = loaves of proposition (0268D turns on proponens), basis = base (0274B: Basis est fundamentum … quasi bases a fundamento surrexerunt), emunctorium = snuffers (must share a root with emuncta), lignum setim = setim wood (0273A says outright the Hebrew name was not translated), sagum = tent-cloth (Douay collapses it into cortina = curtain and erases a distinction CAPUT XXVI spends two columns on), scyphus = goblet. ⛔ This is the 11551 stylus/epistylium class: anglicise one of these and you break the gloss that explains it.
usque ad → as far as, italics kept.

### The Hebrew, and the book-title banner

Migne sets Hebrew type inline at seven places and Corpus Corporum drops every one. All seven were recovered from the plate and are carried into the English verbatim, like [n:] content; Migne's own transliteration beside each is what gets translated. Provenance: data/hebrew-patches/8949.json.
The book opens with a display line under the LIBER EXODUS banner — Hebraice VEELLE SEMOTH (ואלה שמות), id est, HAEC SUNT NOMINA; Graece ΕΞΟΔΟΣ, id est EXITUS. — dropped whole by our source, recovered at the plate on two witnesses, encoded <head type="book-title">. It is not a section head. This is the fifth and last of the Glossa Pentateuch book-title lines; that class is now closed.

### ⛔ The two witnesses — the rule this book settled, with evidence in both directions Gallica is a photograph and it decides. The archive copy is 1-bit JBIG2, a symbol-substitution codec: use it to corroborate a WORD, never to settle a LETTER, and never to clear a defect Gallica shows. Both directions are attested here, one stint apart:

Stint 0005–0009 found four readings that looked like Migne's broken type from our twin alone — guam, apud. Deum, reunquenqus, vasa. — and all four were our own digitization.
Stint 0035–0039 found the archive silently repairing Migne three times in thirteen columns (Sicuttibi 0277A, snpra 0283C, uque ad 0284B — all genuinely on the plate). Trusting its agreement would have withdrawn three real defects; trusting its disagreement would have filed two false patches.
⚠ 0249A is an open disagreement, NOT a press variant. Gallica prints non videant … nos intelligant; the archive prints non vident … non intelligant. Both archive readings move toward the expected Lc 8:10 wording, which is what a symbol-substitution codec does. A third witness (Google Books YfwQAAAAYAAJ) would settle it. Migne's nos is marked [sic:] on Gallica's authority, and the [var:] beside it carries the separate citation question.

### What the plate reading came to 113 of the work's 114 columns were read at Migne's plate, printed corner numbers checked first on every leaf, and every one of the 149 markers stands on a column that was read. Only col 218 is unread and nothing stands on it. 36 TEI patches were filed against our Latin, and every one of them is a defect of Corpus Corporum's digitization, not of Migne — each caught by a translator who opened the plate while the words were still in view, and each of which would otherwise have shipped as a public [sic:] accusation against him.

### The blind polarity read (four fenced readers, all 44 chunks) Six live sites, all new. Two were defects of OURS — the class that leaves no trace and that verify-english cannot see — and both are repaired:

0018 @0230C, a dropped nisi: Non enim fieret nisi quod non erat, "nothing is made except what it was not," is what licenses the anti-Marcionite argument that follows. Our English asserted the contrary. ⚑ Negation-particle counts matched on both sides, which is exactly why this check may never be automated.
0025 @0251B, quia read as causal where it is indirect statement. Paul did not add that likenesses are nothing; our English said they are nothing, and the next four sentences argue the opposite.

Migne's own, marked or logged but never emended: 0026 @0253C [cj: injustus; read justus], settled by Migne's own doublet of the same Origen passage at 0254B; 0021 @0239A Nono die for Novo die; 0007 @0199D a dropped non (logged only — the imperfect dabantur leaves the conjecture arguable); 0038 @0283D aere where Ex 30:3's lemma and the neighbouring gloss on the altar's metal want auro — plate-checked at f142, and the ære is Migne's own, so no patch and no [sic:]; the usque ad abridgment hides the clause that would settle the conjecture.

### ⛔ OPEN — not settled at merge

Migne's foot-of-page note layer exists in this book and we carry none of it. A long editorial note on the division of the Decalogue runs across the foot of f128 onto f129, keyed by a raised letter. Coverage is partial, not a checked zero. Wilson's encoding ruling is owed — the raised-letter key is neither the numbered layer (Latin-only [cn:]) nor the asterisk layer ([cn: * | …] plus a translated [nt:]). See data/briefs/8949-ADJUDICATIONS.md §4.
Three patches held back deliberately: 0186D (divirorum, provisional — Gallica under-inked at the letter), 0219B (a Greek iota needing a second witness), and three markup-only sites where a citation should be a <note> but carries a <pb> inside it. An uncertain read does not move the Latin. ⭐ A fourth, 0295A annuns, was held and then confirmed at 14× and applied — a blind reader had independently flagged it as the one place our English silently repaired Migne, which is precisely the inconsistency the hold created.

Chunks 0000, 0001, 0002, 0003, 0004. Plate range Gallica bpt6k5505319w f92–f97, all six leaves opened, printed corner numbers read first on every one. Apparatus companions: data/briefs/8949-PLATE-READS-0000.json, data/briefs/8949-PENDING-TEI-PATCHES-0000.md.

§1 · Span count against the brief — 94 against 94, EXACT

The brief declares 94 marked spans, of which 1 open-ended, for chunks 0000–0004.

Counted in my own Latin two ways, as the brief requires:

chunk	italic pairs, paragraph-bounded	italic spans, regex	guillemets `«`
0000	7	7	0
0001	22	22	0
0002	20	20	0
0003	27	27	0
0004	17	17	1
total	93	93	1

93 italic + 1 guillemet = 94. The paragraph-bounded and unbounded counts agree, and they agree because no paragraph in this range contains an odd number of asterisks — I checked that rather than inferring it from the totals, since the 8950 0004–0007 precedent is a false reconciliation of exactly this shape. The single « at 0192A is the work's only guillemet and it is unclosed, as the brief states; that is Migne's own setting and was not touched.

§2 · Marker tally

marker	count	columns
`[var:]`	8	0184A, 0185C, 0185D, 0188D, 0189B, 0189D, 0192D, 0193B
`[cj:]`	1	0183A
`[sic:]`	0	—
`[ed:]`	0	—
`[d:]`	0	—

Every marked column was opened at the plate by me. No marker in this stint stands on an unread column, and none stands on the second witness alone.

§3 · `[cj:]` — 0183A, `corporalis` where the argument requires `spiritualis`

Migne prints (plate-confirmed, f92 left column, 2.4× crop):

Per corporalem enim exitum filiorum Israel de Aegypto corporalis noster exitus de Aegypto spirituali signatur.

Rendered faithfully, the English says that our bodily going out from spiritual Egypt is signified — which asserts the opposite of the sentence's own structure. The whole prothema is a catalogue of figures in which a bodily thing signifies a spiritual one, and this very clause pairs Israel's corporalem exitum against de Aegypto spirituali: the second term must be spiritualis. This clears the Pattern 18 bar (the English asserts something the author did not), so the gloss is additive and both readings stand on the page:

our bodily going out [cj: corporalis; read spiritualis, "our spiritual going out"]

The word is a perfectly good Latin form, so there is nothing for a [sic:] to wrap, and the plate is not defective — it is Migne's copy-text, or his compositor, saying something the argument cannot use.

§4 · Two `[sic:]` candidates DECLINED, with the evidence

Both are single-cluster orthographic variants of very common words, both are plate-confirmed as Migne's type, and both were declined under Pattern 9 (attested medieval spellings are correct Latin for their period; marking them buries the real defects in noise).

0184B divictis for devictis. Plate-confirmed verbatim at f92 right column, 3.4×: et divictis eis [devicto eo] discedant de terra sua. The di-/de- prefix alternation is standard medieval orthography. Two further reasons not to mark it: Migne has already flagged the place himself, printing his own bracketed alternative [devicto eo] immediately after, which does the disclosing job our marker would do (the same argument as the "Migne's own ( sic )" rule); and the archive witness prints that bracket as parentheses, which is the JBIG2 copy manufacturing a difference — Gallica is decisive that the brackets are square. Corpus control: divict occurs once corpus-wide (this site) against 30 devict, so it is not house usage; but frequency is not the Pattern 9 test, and the form is attested. Rendered as ordinary English ("these being overcome"), with [devicto eo] carried verbatim as the printed editorial bracket it is.

0185A concientiam for conscientiam. Plate-confirmed at f93 left column, 3.2×, hyphenated across a line as con- / cientiam, and the archive witness reads it the same way. The sc → c simplification is a standard medieval orthographic reduction. Corpus control: concientia occurs twice corpus-wide against 816 conscientia — again not house usage, and again not the test. Rendered as ordinary English ("within the conscience"), no marker.

⚠ Both are borderline and I have recorded the counts so a later reviewer can overturn either without re-doing the plate work. What I was not willing to do was fire a public accusation against Migne's plate on an orthographic variant that his own century set routinely.

§5 · The 7a″ collation — 8 `[var:]`, and every rejection reasoned

Collated against sources/vulgate/clementine-flat.txt, cited by book chapter:verse. Every ⚠ and ⚑ entry in the brief was walked, and so were the ✓ entries whose address I could check.

Fired

0184A · Phlp 1:24. Migne: Permanere autem in carne magis necessarium est propter vos. Clementine: permanere autem in carne, necessarium propter vos. Migne adds the comparative. English shows it ("is more needful"), so the divergence is visible and marked.
0185C · Mt 4:10. Migne: Vade retro, Satana. Clementine: Vade Satana. Plate-confirmed at f93; retro is on the page. The Mt 16:23 / Old Latin form.
0185D · Jr 17:11. Migne: Clamavit perdix, congregavit quae non peperit … in dimidio dierum ejus derelinquent eum, et in novissimis suis erit stultus. Clementine: Perdix fovit quae non peperit … in dimidio dierum suorum derelinquet eas, et in novissimo suo erit insipiens. The most load-bearing divergence in the range: Origen's whole exposition turns on congregavit and on relinquent eum (Intelligit ergo perdicem se … et quod hi, quos sine judicio congregavit, relinquent eum). Plate-confirmed verbatim.
0188D · Ps 44:12. Migne: quia concupivit rex speciem tuam. Clementine: Et concupiscet rex decorem tuum. Perfect for future, and speciem for decorem. The tense matters: the gloss reads it as accomplished in the Church from the Gentiles. Plate-confirmed.
0189B · Rom 10:19. Migne: Ego in non gentem in aemulationem vos inducam … in iram vos concitabo. Clementine: Ego ad aemulationem vos adducam in non gentem … in iram vos mittam. Two lexical substitutions plus reordering. Plate-confirmed.
0189D · Jr 9:25–26. Migne: Considerabo omnes praeputium circumsecantes, Aegyptios et filios Edom. Clementine: visitabo super omnem qui circumcisum habet praeputium, super Aegyptum, et super Juda, et super Edom. A radically different form, and load-bearing: Theodoret's point is precisely that the Egyptians circumcised. Plate-confirmed.
0192D · Ex 3:8. Migne: In terram bonam et spatiosam, in terra quae fluit lacte. Clementine: in terram bonam, et spatiosam, in terram quae fluit lacte et melle. Plate-confirmed that the shift accusative → ablative is really on the page. Marked because English has an exponent for it ("into a good and spacious land, in a land which flows with milk"), so the divergence is visible to a reader rather than silently smoothed.
0193B · Ex 3:22. Migne: et praedabimini eos. Clementine: et spoliabitis Aegyptum. Different verb and different object, in the proof-text Augustine's whole quaestio is about. Plate-confirmed verbatim, italics included.

Rejected, and why — the ⚠ entries that are innocent

0183B · 1Cor 10:3–4 spiritualem for Clementine spiritalem (twice). Orthographic variant of one word; eamdem/eumdem match exactly. Not a divergence.
0183B · Jo 19:36 — Migne's span lacks enim only because Pattern 5's inquit stands in its place (Facta sunt haec, inquit, ut Scriptura impleretur). Not a divergence.
0183A · Jo 6:59 coelo for caelo. Spelling.
0184A · Ex 1:10 Venite, opprimamus eum sapienter against Clementine Venite, sapienter opprimamus eum. Word order only, same words, same sense.
0184A · Ps 13:1 — the gloss recasts Dixit insipiens as a relative clause (qui dixit in corde suo) to make the devil the subject. The gloss's syntax, not the quotation's wording.
0185B · Ps 105:35 — Migne drops only the leading et.
0185C · 2Cor 11:14 transfigurans se (participle) for transfigurat se, to fit the governing clause. Grammatical adaptation.
0187A · Ex 1:17 — Migne's span omits autem obstetrices because his own sentence has just supplied it (dicit Scriptura de his quod …). A fragment, not a variant.
0188D · Ex 2:14 Nunquid me occidere tu vis? against num occidere me tu vis. Particle synonym plus transposition; no sense moves. Plate-confirmed as Migne's.
0190C · Jo 9:28 Tu discipulus illius sis, nos discipuli Mosi against … nos autem Moysi discipuli sumus. Elliptical abridgment of the second clause; sense identical.
0190C · Ex 2:14 lemma Timuit Moses for Timuit Moyses. The name is spelt Moses throughout this work; house usage, established by its own repetitions.
0192A · Jo 1:27 Cujus non sum dignus solvere corrigiam calceamenti against cujus ego non sum dignus ut solvam ejus corrigiam calceamenti. Standard abridgment; and this is the span carrying the book's lone «, which is settled and was not reopened.
0192D · Ps 104:45 — Migne drops one ejus (et legem requirant). Immaterial.
0190D · Ex 2:23 — ⚑ this is the entry the brief's pre-scan flagged as matching only elsewhere, and it earned its flag. Migne prints Post multum vero temporis; the Clementine prints Post multum vero tempore. That single word is why the phrase matched Mt 25:19 and Tob 1:18 and not its own verse. No marker fired, for two reasons: no sense turns on it and English cannot show it ("after much time" either way); and post multum … tempore is not sound Latin, while Migne's partitive genitive multum temporis is — so by the brief's own comparand test the doubtful reading is on our side of the collation. Plate-confirmed that Migne reads temporis. ⚠ Candidate for the comparand-defects list, offered but not asserted: unlike cornonabit, tempore here may well be the Clementine's genuine (if awkward) text rather than a slip in Tweedale's file, so I have not added it to that list myself.

§6 · Readings rendered literally, no marker (Pattern 7 / Pattern 9)

0183A et Hebraeorum liberationum — plate-confirmed genitive plural in a run of accusatives (Per typici agni immolationem, et Hebraeorum liberationum, veri Agni passio). A real form in the wrong case; English has an exponent ("and of the deliverances of the Hebrews"), so rendering it literally leaves the oddity visible. No [sic:] (nothing to carry), no [cj:] (nothing false is asserted).
0192B tanto delectabiliter … quanto — plate-confirmed, comma included. Gregory's text has the comparative delectabilius. Declined as a [cj:]: the correlative still carries in English ("so delightfully … in proportion as"), and no false assertion results. Pattern 18a's census is closed and this is well below its bar.
0190D etc — the plate prints it with no full stop at a line end. Pattern 8: the mark was not supplied, and the English prints etc bare. This will look like a typo to a reader; it is the plate.
0189A tibin — Origen's transliterated θῖβιν, the ark of bulrushes. Kept as printed and left in the English as a technical term the text itself defines two clauses later (Tibin ergo est tegminis genus ex papyro contextum). Not a defect.
0186A / 0191B admistis / admistus — attested syncopated forms of admisc-; used twice in the range, so house usage. Pattern 9.
0188C Salphaat for Salphaad. Proper-name orthography; left as printed.

§7 · Recovered Hebrew and the book-title banner — carried, not touched

Both were settled before this stint and I confirmed each on the plate in passing rather than re-adjudicating it.

Chunk 0000 banner. Confirmed on f92: Hebraice VEELLE SEMOTH ( ואלה שמות ), id est, HAEC SUNT NOMINA; Graece ΕΞΟΔΟΣ, id est EXITUS. is set as a display line above and across the PROTHEMATA head, with LIBER above it. Translated as the display line it is; the Hebrew and the Greek carried verbatim, Migne's transliterations Englished around them.
0183B ואלה שמות and 0193A אהיה carried verbatim into the English, unpointed as Migne sets them, with the spacing of the Latin. The transliterations beside them (Veelle Semoth, ehieh) are what the English renders. No marker fired on either, per the brief.
0183B also carries Migne's Greek ἒξοδος inline; carried verbatim, and the three Latin equivalents beside it (exodus, exitus / egressus) stay Latin because the sentence names them as the Latin words rather than using them.

§8 · Structural verification run before shipping

node scripts/verify-english.mjs 8949 — zero errors on 0000–0004; the only failures reported are the 39 chunks belonging to peer stints, which are not yet written. Additionally checked by hand, per chunk: frontmatter byte-identical to the Latin twin; column anchors identical in count and order; [n:] contents identical in count and order (no note translated to [nt:] in this range — every one is a citation, none is editorial prose); one ## head per chunk, with (cont.) at 0002 left untranslated; guillemet parity 1 « / 0 », matching the Latin exactly. English/Latin word ratios 1.48–1.55, inside the expected ~1.5× band. Em-dashes: 1 in five chunks (0.2/chunk against a corpus median of 5.3).

Stint: CAPUT IV–VII, Moses before Pharao and the first plagues. All seven leaves of the range were read at the plate before anything below was written.

§1 · Span count against the brief — 133 against 133, EXACT

data/briefs/8949-lemmata-0005.txt declares 133 marked spans for chunks 0005–0009. Counted in my own Latin: 133 whole-file, 133 paragraph-bounded — the two agree, which is the check the brief asks for, and the agreement is not the false kind, because this book prints no guillemets at all in my range (0 «, 0 » across all five chunks), so there is no unmatched delimiter that could make two errors cancel. Per chunk: 0005 = 38 · 0006 = 8 · 0007 = 33 · 0008 = 14 · 0009 = 40.

§2 · Marker tally

marker	count	columns
`[var:]`	6	0195A · 0196A · 0196C · 0202B · 0203A · 0203B
`[sic:]`	0	— every candidate resolved to OUR digitization, not Migne's type (§4)
`[cj:]`	0	—
`[d:]`	0	—
`[ed:]`	0	—

Every column carrying a marker was read at the plate: 0195, 0196, 0202, 0203. See data/briefs/8949-PLATE-READS-0005.json.

§3 · The 7a″ collation — 6 `[var:]`, and why each fired

0195A · Ps 72:23. Migne prints Ut jumentum actus sum apud te; the Clementine has ut jumentum factus sum apud te. A real word, a different claim ("I was driven" for "I was made"), inside a quoted scripture — so Pattern 14, not Pattern 18, per that pattern's own boundary rule. The plate reads actus (Gallica f98, native crop, left column).

0196A · Ex 4:12. Augustine's lemma Perge igitur is the Vulgate's; but the text he then quotes and expounds is not — Migne prints Sed nunc vade tu, et ego aperiam os tuum et instruam te quae locuturus es, against the Vulgate's Perge igitur, et ego ero in ore tuo: doceboque te quid loquaris. The whole gloss rests on the divergence: Augustine's point is that the promise is double — "he did not say, Open thy mouth, and I will instruct thee; but he promises both, I will open, and I will instruct" — and neither verb is in the Vulgate wording. Marked.

0196C · Ex 4:19. Migne prints Vade, perge in Aegyptum against the Vulgate's Vade, et revertere in Aegyptum, and Augustine's next sentence turns on it exactly: Primo enim inquit: Vade, perge in Aegyptum, tanquam non sufficeret alterum — "as though one were not enough." The argument is about the two imperatives, so a reader meeting the Vulgate's et revertere would not see what is being discussed. Marked; the plate sets it twice, both confirmed.

0202B · Rom 9:5 — the most consequential of the six. Gregory is distinguishing God said by appellation from God said essentially, and he cites Paul as the proof: quod volens Paulus discernere ait: Ex quibus Christus secundum carnem natus est, qui est Deus benedictus in saecula. Migne's quotation omits super omnia — and two lines later Gregory's own gloss demands it: ut ergo ostenderet Christum naturaliter Deum, non tantum Deum, sed Deum super omnia esse memoravit. As printed, the proof-text does not contain the words the argument says Paul "recorded". Read at the plate (Gallica f101, right column, corner numbers 201/202 read first): Migne genuinely prints it short. Marked.

0203A · Ps 80:11. Migne cites psal. LXXX and prints Aperi os tuum et implebo illud; the psalm reads Dilata os tuum, et implebo illud. This one is worth the note precisely because the same commentary quotes the same verse correctly three times in my own range — Dilata os tuum, et ego implebo illud at 0195B, Dilata os tuum, et ego adimplebo illud at 0196A, and et implebo illud at 0196B — so Aperi here is a departure from the book's own practice, not a house form. Plate-confirmed at f102, left column.

0203B · Gn 3:1. Migne prints Serpens erat prudentior cunctis animalibus against the Vulgate's serpens erat callidior cunctis animantibus terrae. Material, and the gloss is built on the very word: the preceding sentence is Serpens vero pro sapientia vel prudentia poni solet, and the citation is adduced to prove it. With callidior ("more crafty") the proof collapses. Plate-confirmed at f102, left column.

Divergences examined and DECLINED, with the reason

These were all walked against clementine-flat.txt by book chapter:verse, not from recollection.

0193D Jn 3:14 (cited Joan. IV) — Sicut exaltavit Moses serpentem for Et sicut Moyses exaltavit serpentem: word order and name-form only. No gloss turns on it.
0193C Mt 10:16 and 0203B — Estote prudentes sicut serpentes drops ergo. Not material. (Note that at 0203B the prudentes is the Vulgate's own, which is what makes the Gn 3:1 divergence beside it visible as a divergence.)
0194C Rom 11:25 — Caecitas enim ex parte contigit adds enim. Connective only.
0194C Apc 17:15 — Aquas, quas vidisti, populi sunt et gentes compresses the verse (drops ubi meretrix sedet and et linguae) and puts Aquas in the accusative. An abridgment, and Migne's own etc. practice elsewhere shows he abridges freely; the accusative is odd but nothing is asserted by it. Crux only.
0195C Ps 37:14 — Migne reverses the two clauses (sicut mutus… et sicut surdus for tamquam surdus… et sicut mutus) and gives qui non aperuit for non aperiens. Substantial in wording, but the gloss (a man shutting his ear) survives either order intact. Declined deliberately — this is the closest call of the range.
0195C/0195D "Non recipies auditum vanum" (cited Eccli. XLV) — Origen's Old Latin form, against Ex 23:1 Non suscipies vocem mendacii. The gloss's whole point (non dixit, Non audies; sed, Non recipies) works against the Vulgate's suscipies as well as against his own text, and the reference itself is Migne's citation, which is refDisplay business, not ours. ⚑ Note that Migne prints the verse twice, differently on the same column: unitalicized Ne recipias auditum vanum at 0195C, italic Non recipies auditum vanum at 0195D. Both plate-confirmed as printed. No marker; the inconsistency is his and is visible in the facing Latin.
0196C Ex 4:19 second half — mortui sunt omnes drops enim. Not material.
0197B, 0198A–0198B, 0200B, 0200D — the free/conflated New Testament citations (Petra autem erat Christus cited II Cor. X; Quid venisti ante tempus torquere nos conflating Mt 8:29; Scimus te quis sis conflating Mk 1:24/Lk 4:34; Nulla enim societas luci ad tenebras turning 2 Cor 6:14's questions into statements; Cui resistite fortes in fide: ut cito conteratur Satanas joining 1 Pt 5:9 to Rom 16:20). These are the commentary quoting from memory, which is what the scripture policy exists to preserve. Rendered as printed, no markers.
0197A "Alia littera" — Migne himself labels the divergent Exodus 4:24–26 text as another reading. He has disclosed it; we do not mark on top of his disclosure. Same logic as his own ( sic ).
0201B Ex 6:12 — the divergent Ecce ego gracili voce sum is expressly labelled LXX. on the plate. Disclosed, so no marker.
0201B Lk 10:24 (cited Lucae II) — Multi reges et prophetae for multi prophetae et reges. An order swap inside an etc. abridgment; nothing turns on it.
0202B Ex 22:8 — Applica illum ad deos for dominus domus applicabitur ad deos. Gregory's form; the argument needs only the word deos, which both have.
0203A 1 Cor 14:1 — Aemulamini spiritualia dona adds dona. Not material.
0204A Job 41:15 — quasi lapis for tamquam lapis. Synonyms.
0204B Mk 3:27 (cited Marc. XII) and 0205A Mt 23:35 — loose paraphrase/conflation of well-known sayings. Same class as above.
0205B Rom 1:23 — commutaverunt gloriam Dei incorruptibilis for mutaverunt gloriam incorruptibilis Dei. Prefix and word order; the sense is identical and the gloss (the change happening in corde impiorum) is untouched.

§4 · Four `[sic:]` candidates WITHDRAWN at the plate — all four were OURS, not Migne's

This is the finding I would most want a later reader to have. Reading the twin alone, my range looked as though it carried four defects in Migne's type. Every one of them turned out to be the digitization, and each is filed as a pending TEI patch instead.

0195A guam for quam (multo amplius guam apud nos pendens). A non-word; on the twin alone it satisfies Pattern 7's non-word class exactly. The plate reads quam, cleanly, on both witnesses. Withdrawn.
0195A apud. Deum — a stray point mid-phrase in quod erat in principio apud. Deum. Gallica f98 shows a faint low speck between the two words; the archive copy (p. 103, corner numbers 195/196 read first) prints apud Deum with nothing between them. So the twin's period is ours and Gallica's mark is a flaw in that copy. Withdrawn, patch filed. ⚑ Had I gone by Gallica alone I would have carried a stray full stop into the English under Pattern 8.
0198A reunquenqus for relinquendus (reunquenqus est mundus). The strongest-looking [sic:] in the range — a flat non-word in the middle of a plain sentence. Gallica f99 shows a worn, broken impression of a long word; the archive copy at p. 104 reads relinquendus clean. The English renders the plate ("the world must be left behind") per the 2026-09-04 ruling, and the patch is filed.
0204C vasa. et liberandas — a full stop breaking a sentence before et. Both witnesses print a comma: the Gallica mark at 9× magnification carries a descending tail that the true full stops on the same page (e.g. dicens . at 0198D) do not, and the archive prints an unambiguous vasa, et liberandas. Withdrawn, patch filed.

⭐ The class ratio in this range is therefore 0 Migne to 4 ours. That is the 9003 finding repeating, and it is the argument for reading the plate before firing, not after.

§5 · Examined at the plate and deliberately NOT changed

0198D dicens: — ⚠ the two witnesses disagree and I let our twin stand. Gallica f99 shows a single baseline point (dicens . Ne longe abeatis), which is the 8957 @0758A pattern exactly and would have been a patch. But the archive copy prints a clear colon, and this very page of the Gallica copy is the one proved under-inked by the relinquendus case above — a dropped upper dot is precisely that failure. Our colon stands; no patch, no marker. Recorded here so nobody re-opens it from the Gallica image alone.
0201C VERS 20.--, with no period after VERS. The plate really prints it that way, in the same small caps as every other; Migne's own slip. Carried as printed. Contrast the Vers. 9. at 0202D, which is ours — see §6.
0201D "…significat futuros" ends with no terminal stop. Plate-confirmed. Not supplied (Pattern 8).
0201D "ut quisque, de tribu sua uxorem acciperet" — a stray comma after quisque. Plate-confirmed as Migne's. Carried.
0197D (Matth. VI) — Gallica renders this note Matih., which is a named member of the attested Gallica worn-sort list (reference note on 8956, where Matih. was refuted by the archive). Note contents pass through verbatim in any case. No action.
0201C pater excelsus — in Gallica's italic the a reads as u (puter). Inking, not type. No action.
hom. 32 / VERS. 27-31 / hom. 3 — Migne's figure 3 renders like a 5 throughout this fount at scan resolution (VERS. 27-51, hom. 52, hom. 5 in Exod.). It is systematic across every leaf, so it is a rendering artifact and not a reading. Our twin's digits stand.
The foot of every column in the range: CHECKED ZERO, both apparatus layers — no numbered Forte conjectures, no asterisk or letter layer, on any of the seven leaves. This is consistent with the book's recorded spot check and extends it, but per the standing rule it does not license the claim that Exodus has no foot-of-page notes anywhere.

§6 · One case where the plate corrected a label — `Vers. 9.` at 0202D

Our twin prints Vers. 9.-- in lower case where all 22 other verse addresses in my range are VERS. in small caps. That is exactly the shape of a Pattern 9 "section label" question, and I was prepared to normalize it in the English on Pattern 9's authority. The plate settled it instead: Migne sets VERS. 9. in the same small caps as every other, so the lower case is the digitization's. Patch filed; the English reads VERS. 9. ⚑ Worth recording as a method point — Pattern 9 would have produced the right English here for the wrong reason, and left the Latin wrong for ever.

§7 · Readings rendered literally, no marker

0196D Mosi vel infantis uxorem — "the wife of Moses or of the child," which is what Augustine's compressed sentence says and is odd but construable. Rendered as printed.
0202A et tu illi in his quae ad Deum; — the Vulgate's Ex 4:16 ends quae ad Deum pertinent; Migne's quotation stops at Deum. Rendered as the elliptical phrase it is ("and thou to him in those things which are unto God"), because Augustine's argument is precisely about what was and was not said. No [var:]: the ellipsis is an abridgment, not a divergent reading.
0200B Quoniam fetere, — the lemma is cut before its governing verb (Ex 5:21 quia fetere fecistis odorem nostrum). Rendered as the truncated cue it is, "Because to stink, etc." The English is abrupt because the lemma is abrupt.
0204A Apo tou δέρκειν — Migne sets the preposition and article in roman letters and the infinitive in Greek. Carried exactly as printed, Greek included, untranslated; his own gloss id est videre is what gets rendered.
0198B Vides… videbitis — Migne moves from the singular to the plural inside two clauses. Under Pattern 17 the English carries the number: "Thou seest what raw pride does; you will see a little later…" The shift is Origen's and is now visible with no apparatus.

§8 · Conventions aligned with the peer stints

Allegorically. · Mystically. · Morally. · Similarly. (English adverbs), ## CHAPTER N., as far as for usque ad, and the [var: Vulg. <book ch:v> latin, "gloss"] form all match chunks 0000–0004 and 0010–0011, which landed while this range was in progress. One voice within the work.

§9 · Structural verification run before shipping

Column anchors, [n:] notes and ## heads compared 1:1 against each Latin twin, by list and not merely by count: 0005 = 14 anchors / 19 notes · 0006 = 3 / 3 · 0007 = 13 / 20 · 0008 = 4 / 5 · 0009 = 14 / 19. All MATCH in content and order. Guillemets 0 in both columns throughout, as this book requires. node scripts/verify-english.mjs 8949 reports no error against any of 0005–0009 (its failures are the peer chunks not yet on disk). No em-dash was introduced in any of the five.

Stint 0010–0014, 2026-09-08. CAPUT VIII–XII: the remaining plagues, the passover, the exodus. Every column on which a marker stands was opened at the plate (Gallica bpt6k5505319w, leaves f103–f111, printed corner numbers read first on all nine). Plate reads are recorded in data/briefs/8949-PLATE-READS-0010.json; transcription divergences in data/briefs/8949-PENDING-TEI-PATCHES-0010.md.

Span count: the brief declares 165 marked spans for this range. My own Latin counts 165 italic spans, and 165 again when counted paragraph-bounded, with zero guillemets in all five chunks (consistent with the brief's statement that Exodus's single « stands at 0192A, outside my range). The totals agree.

1. ⭐ THE HEADLINE FINDING — Rabanus's Decalogue is not quoted from Exodus 20

Rabanus's running series of precepts (secundum … tertium … quartum … quintum … sextum … septimum … octavum … nonum … decimum praeceptum) is the spine of this whole range: each commandment is matched to its answering plague. Migne's wording for seven of the ten does not agree with the Decalogue as Exodus 20 prints it, and two of them agree instead with a different book. Collated against sources/vulgate/clementine-flat.txt:

col	Migne prints	Ex 20	verdict
0205D	Non accipies nomen Dei tui in vanum	20:7 Non assumes nomen Domini* Dei tui in vanum*	`[var:]` — verb changed, Domini dropped
0207A	Memento diem sabbati sanctificare	20:8 Memento ut diem sabbati sanctifices	`[var:]` — infinitive for ut + subjunctive
0207D	Honora patrem tuum et matrem tuam	20:12, exact	agrees; no marker
0209D	Non mœchaberis (plate)	20:14 Non moechaberis, exact	agrees; see §3
0210B	Non occides	20:13, exact	agrees; no marker
0211A	Non furaberis	20:15 Non furtum facies	`[var:]` — and see below
0214C	Falsum testimonium non dices	20:16 Non loqueris contra proximum tuum falsum testimonium	`[var:]`
0215D	Non concupisces uxorem proximi tui	20:17 Non concupisces domum* proximi tui, nec desiderabis uxorem ejus*	`[var:]` — and see below
0220A	Non concupisces rem proximi tui	20:17 (as above)	`[var:]`

⭐ Two of these are the brief's own warned case — a ✓ against the wrong address. The pre-scan marked Non furaberis ✓ Rom 13:9 and Non concupisces uxorem proximi tui ✓ Dt 5:21. Both ticks are real and both are the finding: Rabanus is enumerating Exodus's Decalogue, and Migne's wording at those two places is Romans 13:9's and Deuteronomy 5:21's respectively. Non furtum facies (Ex 20:15) and furtumque non facies (Dt 5:19) are the two forms the Decalogue actually carries; Non furaberis is neither, and it is Paul's.

Every one of these was read at the plate (f103 c1, f104 c1/c2, f105 c1, f106 c1, f107 c2, f108 c1, f110 c2) and is Migne's setting, not a transcription artifact. The class looks like a catechetical short form of the Decalogue circulating in Rabanus's source rather than a series of compositor's slips: the divergences are consistent in kind (abbreviation, and NT phrasing), and the two that agree with Exodus agree with it exactly.

2. `[var:]` markers standing outside the Decalogue class

0207A — Isaiah 66:2, and the gloss rests on the word Migne prints. Migne: Super quem requiescet Spiritus meus nisi super humilem et quietum? The Clementine has nothing of the sort — ad quem autem respiciam, nisi ad pauperculum, et contritum spiritu, et trementem sermones meos: a different verb and different objects. The exposition immediately following turns on quietum: Inquieti ergo a Spiritu sancto resiliunt … nec admittunt quietem sabbati spiritualis. Conform the lemma to the Vulgate and the argument loses its keyword. Plate-confirmed on f104 c1 (col 207, band A). This is 7a″'s founding shape exactly.

0210D — Romans 9:20, and Migne prints the verse twice, differently. He sets O homo, tu qui es qui respondeas Deo (Rom. IX), which is not a question and has no main predicate; eleven lines below, at 0211A, he sets the same verse correctly as O homo, quis es qui respondeas Deo. Both readings were read at the plate on f105 c2 (band D) and f106 c1 (band A) — two separate columns, two separate crops. Rendered literally the page reads "O man, thou who art, who repliest against God", which loses Paul's rebuke; the [var:] carries the received interrogative beside it. [sic:] does not reach this (qui is a real word) and Pattern 18 [cj:] is explicitly excluded inside a quoted scripture, so Pattern 14 is the right home. ⚠ Under 7a⁗ question 3, a swallowed interrogative converts a question into a claim: the English therefore does not quietly restore the question, and the note is what discloses it.

0214C/D — Galatians 5:15, and again the gloss follows Migne. Migne prints Si mordetis et consumitis, videte ne ab invicem consumamini (Gal. v) where the Vulgate has comeditis. Two lines above, Rabanus's own gloss reads Falsus enim testis nocet mordendo et consumit mentiendo — the exposition's verb is Migne's, not the Vulgate's. Plate-confirmed on f107 c2 (col 214, bands C–D).

0217A — Isaiah 7:15, whose second clause is wholly different. Migne: Butyrum et mel manducabit; priusquam faciat aut proferat maligna, eliget bonum (Isa. VII). The Clementine: Butyrum et mel comedet, ut sciat reprobare malum, et eligere bonum. Origen's own continuation picks up Migne's word — quoniam priusquam sciat puer bonum aut malum — so again the argument stands on the printed reading. Plate-confirmed on f109 c1 (col 217, band A).

3. ⛔ CANDIDATES WITHDRAWN AT THE PLATE — three, and each would have been a false claim

These are recorded so nobody re-opens them from the Latin alone. All three look like defects in our TEI twin and are not.

(a) 0209D maechaberis — the plate prints the œ ligature. Our twin reads Non maechaberis, which against Ex 20:14's moechaberis is a live divergence and, on the ae/oe axis, a live Pattern 9 argument. The plate (f105 c1, band D, native-resolution crop) reads Non mœchaberis — Migne's own ligature, which our transcription has read as ae. So there is no divergence from the Vulgate at all: Migne agrees with Ex 20:14 exactly. The English renders the plate ("Thou shalt not commit adultery"), no [var:] and no [sic:] fires, and the TEI is filed for patching. ⚑ The same page prints quaedam mœchia at 0215D, where our twin correctly has oe — so the misreading is local to one sort, not systematic.

(b) 0207A circum venit — a word broken at a line end, with no hyphen. Our twin sets et tanta calliditate circum venit, ut deceptus nec videat… — a split run that satisfies Pattern 10 exactly and would have taken a [sic: circum venit]. The plate (f104 c1, band A) breaks circum | venit across a line end with no hyphen: it is one word, circumvenit. This is the class CLAUDE.md names as "ours and it reads as Migne's", and the 8957 Per eat/Pereat precedent is the same shape. Marker withdrawn; rendered "encompasses".

(c) 0214D [n: (ORIG.)] hom. 4 in Exod.) — the unbalanced parenthesis is MIGNE'S. Our twin prints an italic run containing a closed (ORIG.) followed by hom. 4 in Exod.) with a second closing parenthesis and no opener — which reads unmistakably as a Corpus Corporum markup failure, and I had it drafted as a TEI patch. The plate (f107 c2, band D) sets it exactly so: (ORIG.) hom. 4 in Exod.). Migne's own defective punctuation. Pattern 8 governs: preserve, never repair. No patch, no marker; the citation rides untranslated as printed.

4. Candidates examined and DECLINED (with the reason, so they are not re-raised)

0208B Digitus Dei hic est, word order. Migne inverts est hic here alone; the same clause prints Digitus Dei est hic five other times in my range (0207B, 0207C, 0212A, 0212D, 0217A) and Ex 8:19 has est hic. Plate-confirmed as Migne's (f104 c2, band B). Both readings are sound Latin and the sense is identical; English has no exponent for Latin word order, so rendering the two differently would invent a distinction Migne does not make. Both render "This is the finger of God" and the divergence is recorded here. Same logic as Pattern 9's gender rule.
0210D cui vult miseretur for Rom 9:18 cujus vult. Plate-confirmed. Misereor takes the dative in late Latin as readily as the genitive; the referent and the sense are unchanged, and no agency is mis-bound. No marker.
0212C Ignem veni mittere in terra for Lc 12:49 in terram. Plate-confirmed (f106 c2, band C). A single letter, and the ablative for the accusative of goal; English "in the earth" carries the printed case. Below the Pattern 14 bar (no sense change), and a single-letter call is exactly the class the 9004 measurement warns against. Crux only.
0212C, the doubled Luke reference. Migne prints de quo dicit Dominus Luc. XII : in the running text and (Luc. XII) as a parenthetical after the quotation. Plate-confirmed. This is his ordinary citing habit (attribution in prose plus locator note), not dittography — Pattern 11 concerns a repeated run of text. No [d:].
0207C Jamnes for Jannes, and mente corrupti for 2 Tim 3:8 corrupti mente. Both plate-confirmed (f104 c1, band C). Jamnes is an attested form of the name; the inversion changes nothing and nothing in the gloss depends on it. Also Sicut opens the quotation where the Vulgate has Quemadmodum autem, which is the ordinary trimming of a quotation introduced by Unde.
0217D Nisam for Nisan. Plate-confirmed (f109 c1, band D), with Hic est Nisan printed eight lines below in the next gloss. Tempting, and declined: it is a proper name, so there is no lexicon against which it fails, and the 8950 Bethehem/Bethlehem precedent is directly on point — crux only, no [sic:], no patch.
0205D qui loquitur mendacium, de suo loquitur (Jo 8:44 cum … ex propriis). Migne's de suo is what makes Rabanus's antithesis work (de Deo loquitur … de suo loquitur); it is an adaptation, carries no citation, and the sense is identical. Declined.
0206A Fecerunt autem malefici (Ex 8:7 Fecerunt autem et malefici). One connective, in an abridged lemma closing with etc. Plate-confirmed. Declined.
0215A Moses for Moyses in the lemma of Ex 10:13. Migne spells the name both ways throughout the range (Moyses at 0210A, Moses at 0211C and here). His own variance; no marker.
0218D poenitus for penitus. An oe/e spelling, Pattern 9. No marker.
Origen's and Augustine's scripture is quoted from their own Old Latin throughout (0210C Et induravit Dominus cor Pharaonis regis Aegypti et servorum ejus; 0212C–D the whole sequence Orate pro me Dominum, Euntes immolate, Peccavi ante Dominum Deum vestrum, Ite et servite, Surgite, exite de populo meo; 0213A Gen 31:42 nunc me inanem dimisisses for forsitan modo nudum me; 0215B Ex 10:12; 0219D Ex 12:22 ex sanguine qui est juxta ostium; 0219B Augustine's aeternum vel aeternalem). Marking each would put a [var:] on every patristic quotation in the book. The class is recorded here once; no markers fired. Where a father's citation is instead the thing his argument turns on, it is treated above (0219B, §5).

5. Migne's mis-bound references — his, and therefore untouched

[n: …] contents are Migne's citations and ride verbatim on both sides, so none of these takes a marker. They are recorded because a reader will otherwise think the fault is ours.

0219C — [n: (Gal. III)] on a quotation of Romans 6:3. Migne prints Quicunque baptizati sumus in Christo Jesu, in mortem ipsius baptizati sumus, etc. (Gal. III). That is Rom 6:3; Gal 3:27 reads Quicumque enim in Christo baptizati estis, Christum induistis. Plate-confirmed (f110 c1, band C). ⚑ Migne's in mortem (accusative) against the Clementine's in morte is a divergence that runs toward the Greek εἰς τὸν θάνατον, so it degrades nothing and takes no [var:].
0221A — [n: (Matth. VI)] on a quotation of Matthew 19:21. Plate-confirmed (f111 c1, band A). The cause is visible on the page: two lines above, the gloss alludes to Mt 6:19–20 (ne thesaurizes ubi tinea exterminat, et fures effodiunt), and the Mt 6 locator has migrated onto the Mt 19 quotation that follows. ⚑ The quotation also reads vade et vende omnia quae habes where Mt 19:21 has vade, vende quae habes — the omnia is Luke 18:22's. Conflation, not a sense-damaging divergence; recorded, no marker.
0218B — [n: (I Petr. II)] on pro peccato mortuus est, cum peccatum non haberet. Nearer 1 Pet 3:18 with 2 Cor 5:21 than 1 Pet 2:22. Migne's citation; recorded only.

6. Renderings decided, and why

⭐ ingravare = "make heavy", indurare = "harden", kept apart throughout. This is not a stylistic choice: at 0209B Augustine's whole point is the difference between Ingravatum est cor Pharaonis and ingravavit Pharao cor suum, and at 0210C–0213B Origen builds an eight-plague argument on induratum esse versus indurasse Dominum, closing at 0213B where Augustine glosses Ego enim induravi cor ejus with dicitur, ingravavi cor ejus. Collapsing both into "hardened" — which is what Douay does at Ex 8:32 — would make three separate expositions unintelligible. Applied uniformly across 0010–0014.
sciniphes → "sciniphs" (the received English of the Douay at Ex 8:16), not italic Latin: an English exponent exists, so Pattern 7's carry-through does not apply.
cynomyia kept in italic Latin, with "dog-fly" beside it where Migne himself glosses it (id est, canina musca). The word is the Septuagint's and the gloss at 0208A is explicitly about the Seventy's choice of it, so an English-only rendering would erase what is being discussed.
calamita, rubeta rendered "the calamite", "the bramble-frog". Migne's own etymology (eo quod in vepribus agat) is preserved by the English form; the plate sets both roman.
0210B Neque enim nihil. Rendered "For neither is it nothing." — both negations carried, per 7a, though the idiom is a litotes.
0216D Nec non locustarum ab ea subducit exempla. Rendered "Nor does it fail to draw off…" — the doubled negative carried rather than flattened into "and also".
Pattern 17 applied throughout: thou/thee/thy for every singular address (0207A–B Esto ergo mansuetus… vacare non vis; 0209D si pecus es; 0210B Si posses videre; 0211A Dum furaris; 0220B Cum laudas… puniris), you only where the Latin is plural (0212D Peccavi ante Dominum Deum vestrum et in vos).
0219C anchor placement. Our twin sets [0219C] inside Christi mortem [0219C] figurante. English word order cannot hold the anchor between those two words; it is placed at the nearest construction boundary ("…begun in baptism, [0219C] that figures the death of Christ"), in order and in the same sentence.
0215C Posuit enim Deus tenebras latibulum suum. Two italic runs with roman enim Deus between them, which English word order cannot reproduce; rendered as two adjacent italic runs after "For God". Both runs preserved, in order.
0219D — a supplied question mark removed. A first draft closed Augustine's third alternative (An, quod est credibilius, ideo dixit…) with a question mark, matching the parallel An praecipit … occidatur? above it. The plate ends it with a full stop (f110 c2, band A), so it stands as a statement opening with "Or" — Pattern 8, never supply. Question-mark counts are now exact against the twin in all five chunks (2/2, 4/4, 0/0, 1/1, 3/3).

7. Two observations that need a witness we do not have

0219B, the Greek. Our twin prints αἰὼνιον. The plate (f110 c1, band B, at 6× native) prints what reads as α ὠνιον — an alpha, a visible gap, then a smooth breathing and grave on the omega, with no iota visible. The word is certainly αἰώνιον (it is what Augustine's quaestio is about). Whether Migne's iota is absent, or worn, or lost in the Gallica photograph, cannot be settled from one witness, and the archive copy is 1-bit JBIG2 and may never be sole witness for a letter. No marker, no patch; filed in the pending-patches file as needing a second witness. The English carries the twin's reading verbatim.
0215D, terminal stop. The last words of chunk 0012 (qui hoc facit vere caecatur furore horribili) show no full stop in the Gallica crop, where our twin has one. It is a single point at the foot of a column and may simply have failed to ink. Not acted on in either direction; the English follows the twin.

8. ⛔ TWO ITEMS FOR MERGE-TIME RECONCILIATION — they cross stint boundaries

Neither of these is mine to settle alone, and both are invisible to verify-english.

8a. praeceptum is being rendered two ways inside one enumerated series

Rabanus's Decalogue series begins in chunk 0009 and ends in chunk 0014, so it is split across two stints, and the two halves do not agree:

chunk	stint	praeceptum rendered
0000	peer	precept (1)
0002	peer	commandment (3)
0009	peer	commandment (3) — Primum praeceptum de colendo uno Deo
0010–0014	this stint	precept (14)

As it stands the reader meets "the first commandment is concerning the worship of one God" at 0205B and then, three lines later on the facing column, "the second precept" — the same series, the same word, two Englishes. That is drift of exactly the kind the merge step exists to catch (11551's stylus, 11063's sense labels, 10804's amica mea).

Recommendation, and the reason is not taste: render praeceptum = precept and mandatum = commandment, because Migne's Latin uses both and one sentence in my range needs them apart — 0206C legem significat, quae mundum decem plagis mandatorum corripit stands eleven lines from 0205D Secundum praeceptum. Collapsing both into "commandment", which is what 0002 and 0009 do, loses a distinction the plate makes. ⚠ This is a recommendation to the merge, not a unilateral call, and I have not edited a peer's chunk. praeceptum occurs in 22 of the 44 chunks, most of them still untranslated, so whichever way it is settled should be settled once and carried forward.

8b. The first precept diverges too, and it is in a peer's chunk

The Decalogue divergence class in §1 above is incomplete as it stands, because its first member falls just outside this stint's range. At 0205B (chunk 0009) Migne prints Primum praeceptum de colendo uno Deo: Non erunt tibi dii alii praeter me, where Ex 20:3 reads Non habebis deos alienos coram me — a different verb and a different phrase, the same shape as the seven divergences in my table. Chunk 0009 fires no [var:] there, which is entirely reasonable from inside that stint: the pattern is only visible once several precepts stand together, and the 0009 stint had exactly one of them.

No action taken here — 0205B is a peer's column and, on the ratchet discipline, firing a marker on a column I read only for its band-C/D content would be the wrong kind of tidying. It is recorded so that the merge can decide whether the class should be marked uniformly across the two chunks or left to the crux, and so that the class is not later reported as a fresh discovery.

Stint: CAPUT XII (cont.) – CAPUT XV — the departure from Ramesse, the Red Sea, and the canticle of Moses. All seven leaves of the range (Gallica f111–f117 = cols 221–234) were read at the plate, both columns, corner numbers first, before anything below was written. No column in this range had been read during prep.

§1 · Span count against the brief — 143 against 143, EXACT

data/briefs/8949-lemmata-0015.txt declares 143 marked spans for chunks 0015–0019. Counted in my own Latin: 143 line-bounded, 143 paragraph-bounded — the two agree. The agreement is not the false kind the brief warns of: this range contains 0 « and 0 », so there is no unmatched delimiter that could let two counting errors cancel. Per chunk: 0015 = 28 · 0016 = 31 · 0017 = 34 · 0018 = 35 · 0019 = 15.

§2 · Marker tally

marker	count	columns
`[var:]`	33	0221C · 0223A · 0223D ×2 · 0224B ×2 · 0224C · 0224D · 0225B ×2 · 0225C · 0226A · 0227A ×2 · 0227C · 0227D ×2 · 0228A · 0228B · 0228C · 0229C · 0230A · 0230B · 0230C · 0230D ×2 · 0231B · 0231D · 0232A · 0233B · 0234A ×2 · 0234D
`[sic:]`	1	0225A (`ververba`)
`[cj:]`	0	—
`[d:]`	0	—
`[ed:]`	0	—

Every column carrying a marker was read at the plate. See data/briefs/8949-PLATE-READS-0015.json.

Four further candidate defects proved to be ours, not Migne's, and are filed as pending TEI patches (data/briefs/8949-PENDING-TEI-PATCHES-0015.md): miqui, id est., scili cet, 6 n Exod.

§3 · The one `[sic:]` — 0225A, `ververba`

Our twin prints Quomodo autem sunt accipienda haec ververba, [0225A] cum viderint postea Israelitae Aegyptios? — and this one is Migne's, not ours, which is the opposite of what this book's four previous candidates turned out to be.

Gallica f112, printed corners 223 / 224, right column, last line of 0224D: Quomodo autem sunt accipienda haec ver- — hyphenated at the foot of the column.
Gallica f113, corners 225 / 226, left column, first line of 0225A: verba, cum viderint postea Israelitae Aegyptios? An.
The archive copy agrees independently (patrologiaecurs04migngoog_djvu.txt ll. 17226 / 17238: hæc ver- … Terba, cam viderint).

So the compositor set the catch-fragment ver- at the foot of one column and then reset the whole word at the head of the next. Read through, the plate says haec ververba. Two independent witnesses, both clean, both showing the doubling; it is not the class CLAUDE.md warns about (a hyphen at a break that our transcription lost), because here our transcription has exactly what both plates print.

Marked [sic: ververba] rather than [d:] deliberately: Pattern 11 wraps a rendered intrusive copy, and the fragment ver has no English exponent to wrap — there is nothing on the English page that would read as our repetition. Pattern 12, as amended 2026-09-05, is the right home: the sense is rendered (these words) and Migne's type sits inside the marker. Conjecture: read verba.

§4 · The 7a″ collation — 33 `[var:]`, and why each fired

Grouped by column. Every reading below was confirmed on the Gallica plate before the marker was fired, and every Clementine reading is quoted from sources/vulgate/clementine-flat.txt by address.

0221C · Os 6:3. Migne: Resuscitabit nos Deus post biduum, et in die tertia resurgemus, et vivemus in conspectu ejus. Clementine: Vivificabit nos post duos dies; in die tertia suscitabit nos, et vivemus in conspectu ejus. Three departures in one clause (Resuscitabit … Deus, post biduum, resurgemus for suscitabit nos), and the gloss is built on the third: the resurrection of the people, not God's raising of them, is what carries the three-day figure into baptism in the next sentence.

0223A–B · Eph 5:8. Migne: Fuistis aliquando tenebrae, nunc autem lux in Domino. Clementine: Eratis enim aliquando tenebrae : nunc autem lux in Domino. Perfect for imperfect, and the connective dropped.

0223D · Mt 7:14. Migne: Arcta est enim et angusta via quae ducit ad vitam. Clementine: Quam angusta porta, et arcta via est, quae ducit ad vitam. Migne's form has lost the gate entirely and turned an exclamation into a statement — and the gloss immediately reasons from ascenditur anguste et difficulter, i.e. from the road, not the gate.

0223D · Lc 14:28. Migne: Quis vestrum volens turrim aedificare, non prius sedens computat sumptus, si habet unde perficiat? Clementine: Quis enim ex vobis volens turrim aedificare, non prius sedens computat sumptus, qui necessarii sunt, si habeat ad perficiendum.

0224B · 2 Tim 3:12. Migne: Qui enim volunt pie vivere in Christo, persecutionem patiuntur. Clementine: Et omnes, qui pie volunt vivere in Christo Jesu, persecutionem patientur. Present for future, omnes dropped, Jesu dropped.

0224B · Ex 14:19–20. Origen's text as Migne prints it: Levavit … angelus Domini, qui antecedebat castra Israel, et abiit post illos. Elevavit et columnam nubis a facie eorum, et stetit post illos, et intravit inter castra Aegyptiorum et Israel. Clementine: Tollensque se angelus Dei, qui praecedebat castra Israel, abiit post eos: et cum eo pariter columna nubis, priora dimittens, post tergum stetit, inter castra Aegyptiorum et castra Israel. In the Vulgate the pillar goes with the angel; in Migne's text the angel lifts the pillar and the pillar itself enters between the camps — which is the whole of the gloss that follows (Haec columna nubis populo Dei efficitur murus). The lemma word Levavit is italicized separately by Migne precisely because it is not the Vulgate's Tollensque se.

0224C · Ex 14:11. Migne: Tanquam non essent sepulcra in Aegypto, ita eduxisti nos. Clementine: Forsitan non erant sepulchra in Aegypto, ideo tulisti nos ut moreremur in solitudine. ⚑ Note that Migne prints the Vulgate form himself six lines later as the verse lemma (Forsitan non erant sepulcra), so the two forms stand on the same column: this is a quotation from a different text, not a compositor's slip.

0224D · Ex 14:13. Augustine's text: Sicut enim vidistis Aegyptios hodie, non apponetis amplius videre eos in aeternum tempus. Clementine: Aegyptios enim, quos nunc videtis, nequaquam ultra videbitis usque in sempiternum. Again the divergence is load-bearing: Augustine's whole quaestio turns on sicut … hodie and on in aeternum tempus, and he says so (An, non eos videbitis sicut hodie … ut nulla sit quaestio nec de aeterno tempore quod hic posuit). Neither phrase exists in the Vulgate wording.

0225B · Gal 4:6. Migne: Quia dabit Deus Spiritum Filii sui in cordibus nostris clamantem: Abba, Pater. Clementine: Quoniam autem estis filii, misit Deus Spiritum Filii sui in corda vestra, clamantem: Abba, Pater. Future for perfect, first person for second.

0225B · Rom 8:26. Migne: ipse Spiritus interpellat pro nobis gemitibus inenarrabilibus. Clementine: sed ipse Spiritus postulat pro nobis gemitibus inenarrabilibus. ⚑ interpellat is the Old-Latin/Vetus reading and is a real verb, so this is Pattern 14 and not Pattern 18; the following sentence (postulante Spiritu sancto) shows the gloss knows postulat as well.

0225C · Ps 67:28. Migne: Ibi Benjamin minor in stupore. Principes Juda, duces eorum, principes Zabulon et principes Nephthali. Clementine: Ibi Benjamin adolescentulus, in mentis excessu; principes Juda, duces eorum; principes Zabulon, principes Nephthali.

0226A · Eph 6:16. Migne: Assumentes scutum fidei, ut possitis omnia maligni ignita jacula exstinguere. Clementine: in omnibus sumentes scutum fidei, in quo possitis omnia tela nequissimi ignea extinguere. ⚠ Migne does not italicize this one, though he introduces it with secundum illud Eph. VI. Marked anyway: Pattern 14 attaches to a scripture citation, not to a typographic class.

0227A · Jo 17:1. Migne: Pater, venit hora, glorifica Filium tuum. Clementine: Pater, venit hora: clarifica Filium tuum. ⭐ This one is the most consequential in the stint. The entire gloss is a distinction between glorificari and gloriose glorificari, and it is built on the verb: the Vulgate's clarifica would break the argument outright.

0227A · Lc 24:26. Migne: Quia oportebat pati Christum, et sic intrare in gloriam suam. Clementine: Nonne haec oportuit pati Christum, et ita intrare in gloriam suam? A question turned into a statement, which is what the argument needs it to be.

0227C · Hab 3:8. Migne: Equitatus tuus salus. Clementine: Qui ascendes super equos tuos, et quadrigae tuae salvatio. Migne's equitatus is exactly the word the gloss then works (de equitatu salutis erat); the Vulgate's quadrigae would not serve, and Origen's horse-and-rider figure needs the cavalry, not the chariots.

0227D · Jo 16:28. Migne: Vado ad Patrem meum. Clementine (Jo 16:28): vado ad Patrem, without meum. Marked because meum is the pivot: the gloss reads Deus meus … Deus patris mei off it.

0227D · Am 9:6. Migne cites Amos IX and prints Dominus nomen est illi. Clementine Am 9:6 reads Dominus nomen ejus; the wording Migne prints is that of Jdt 16:3 (Dominus conterens bella, Dominus nomen est illi) — the very verse whose first half he has just given as his Alia littera. So the pre-scan's bare ✓ against Jdt 16:3 is, read against the address Migne himself prints, a finding. ⛔ The Alia littera itself (Dominus conteret bella) takes no marker: Migne has disclosed the divergence in his own voice, and the standing rule is not to stack ours on his.

0228A · Ex 15:4. Migne: Quadrigas Pharaonis et exercitum ejus projecit in mare: electos ascensores, ternistratores demersit in mare Rubrum. Clementine: currus Pharaonis et exercitum ejus projecit in mare: electi principes ejus submersi sunt in mari Rubro. Unitalicized, but quoted; and ternistratores (τριστάτας) is the word two whole paragraphs then expound. The Vulgate has no such word.

0228B · 1 Cor 3:12. Migne: Qui aedificat supra fundamentum argentum, aurum, lapides pretiosos. Clementine: Si quis autem superaedificat super fundamentum hoc, aurum, argentum, lapides pretiosos. Note the order silver-then-gold, against the Vulgate's gold-then-silver.

0228C · Ex 15:10. Migne: Demersi sunt quasi plumbum in aqua validissima. Clementine: submersi sunt quasi plumbum in aquis vehementibus. Migne's singular aqua validissima is repeated verbatim at 0229A as the clincher of the whole lead-and-gold argument, so the form is the author's, not a slip.

0229C · Ex 15:12. Migne: Extendisti dexteram tuam et devoravit eos terra. Clementine: Extendisti manum tuam, et devoravit eos terra. ⭐ The gloss exists because of this word: quasi ut devorarentur a terra, causa fuerit quod dexteram extendit — and it then reads the right hand as the outstretched arm on the cross. With the Vulgate's manum there is no crux to solve.

0230A · Ex 15:14. Migne: Audierunt gentes et iratae sunt, dolores apprehenderunt habitatores Philisthiim. Clementine: Ascenderunt populi, et irati sunt: dolores obtinuerunt habitatores Philisthiim. The lemma Migne prints two words earlier is Ascenderunt; the text he then quotes says Audierunt. Both stand as printed.

0230B · Is 65:2. Migne cites Isa. LXV and prints Tota die expandi manus meas ad populum non credentem. Clementine Is 65:2 reads Expandi manus meas tota die ad populum incredulum; the wording printed is verbatim Rom 10:21. Another ✓-against-the-wrong-verse.

0230C · Rom 11:25–26. Migne: Caecitas enim ex parte contigit in Israel, donec plenitudo gentium intraret, et tunc Israel salvus fieret. Clementine: quia caecitas ex parte contigit in Israel, donec plenitudo gentium intraret, et sic omnis Israel salvus fieret. tunc for sic and omnis dropped — and tunc is what makes the verse answer the donec pertranseat populus tuus the gloss is expounding.

0230D · Dt 32:6. Migne: Nonne hic ipse Dominus Deus tuus, qui fecit et creavit te, et acquisivit te? Clementine: numquid non ipse est pater tuus, qui possedit te, et fecit, et creavit te? Migne's acquisivit is the whole point of the passage (quomodo acquisisse dicitur quod suum esse dubium non est?); the Vulgate's possedit would dissolve the question.

0230D · Is 50:1. Migne: Peccatis vestris venundati estis, et pro iniquitatibus vestris dimisi matrem vestram. Clementine: Ecce in iniquitatibus vestris venditi estis, et in sceleribus vestris dimisi matrem vestram. The two nouns have exchanged places and venundati stands for venditi — and venundati is the word the gloss repeats twice in the next four lines.

0231B · Ps 79:10. Migne: … et plantasti radices ejus, et replevit terram. Clementine: plantasti radices ejus, et implevit terram. One prefix; recorded because the surrounding three verses are quoted verbatim, which makes the single departure legible rather than noise.

0231D · Ex 15:17. Migne prints the Vulgate lemma Sanctuarium tuum and then Origen's text Sanctimonium tuum, Domine — for which the Clementine has sanctuarium tuum, Domine. Marked on the second, since the gloss then etymologizes Sanctimonium for eight lines (Sanctimonium dicitur tabernaculum Dei).

0232A · Ex 15:18. Migne: Domine, qui regnas in saeculum et in saeculum et adhuc. Clementine: Dominus regnabit in aeternum et ultra. ⭐ The gloss is a word-by-word exposition of in saeculum … et adhuc running a full column, and et adhuc is quoted five further times. Nothing of it survives against in aeternum et ultra.

0233B · Pr 3:18. Migne: Sapientia lignum vitae est in omnibus amplectentibus eam. Clementine: Lignum vitae est his qui apprehenderint eam, et qui tenuerit eam beatus. Migne supplies the subject Sapientia, which is what lets Origen call the wood cast into Mara lignum sapientiae Christi.

0234A · Ex 15:25. Migne: posuit illi Deus justitias et judicia, ibi tentabat eum dicens: Si audieris vocem Domini Dei tui. Clementine: ibi constituit ei praecepta, atque judicia, et ibi tentavit eum, dicens: Si audieris vocem Domini Dei tui. Note that Migne prints the Vulgate form as his own lemma (Ibi constituit ei praecepta) immediately above, so the two readings face each other on the column.

0234A · Ez 20:25. Migne: Dedi vobis praecepta et justificationes non bonas in quibus non vivent. Clementine: Ergo et ego dedi eis praecepta non bona, et judicia in quibus non vivent. Second person for third, justificationes for judicia — and the gloss's next sentence works justificationes.

0234D · Is 7:9. Migne: nisi crediderint, non intelligent. Clementine: Si non credideritis, non permanebitis. The familiar Old-Latin/LXX form; the argument is about understanding the wood, and non permanebitis would not close the chapter.

§5 · Divergences examined and DECLINED, with the reason

Recorded so that a later pass does not re-open them. All were checked against sources/vulgate/clementine-flat.txt and confirmed on the plate.

0221D · Ex 12:46, the lemma Nec illius. Ex 12:46 reads nec os illius confringetis; Migne's lemma cuts to Nec illius, which is not a contiguous run of the verse. This is Migne's own catchword abridgment (he closes it with etc.), not a textual reading: the gloss supplies the missing word in its very first phrase, Os Christi in cruce non est confractum. No [var:].
0224C · Ex 14:11 lemma Forsitan non erant sepulcra. Only sepulcra for sepulchra — Pattern 9 orthography, and the same spelling stands three lines above in Migne's own prose.
0224C · the citation Matth. XIII for Mt 14:31. Migne's printed reference is wrong (the verse is Mt 14:31, and Migne prints Matth. XIV correctly for the same verse at 0228D). Confirmed at the plate, f112 right column. This is a refDisplay matter and is left exactly as printed; no marker.
0233C · the citation II Cor. III for 2 Cor 4:7. Same class, and it is inside an [n:], whose contents are verbatim. The quotation itself (Habemus thesaurum istum in vasis fictilibus) differs from the Clementine only by the dropped autem — an abridgment, no marker.
0230C · Mt 3:9 and 0231C · Mt 3:10. Potens est enim Deus … and Nam omnis arbor … differ from the Clementine only in the connective (enim for Dico enim vobis quoniam; Nam for ergo). No sense turns on either.
0231D · 1 Cor 3:9. Dei agricultura, Dei aedificatio estis against Dei agricultura estis, Dei aedificatio estis — a single dropped estis, an abridgment of a doubled verb.
0232B · Ex 15:19. filius es Israel, potes ambulare per siccum in medio maris is Origen turning the verse into second-person application, not quoting it; italicized as an echo. Not a claim about the text of Exodus, so not a [var:].
0232D · Ex 15:22, Tulit autem Moses Israel. Moses for Moyses is this work's house spelling throughout (it stands so in the running prose at 0221C, 0225B, 0233A and elsewhere). Pattern 9.
0233A · Ex 15:25, ostendit illi lignum Deus. illi for ei; the Vulgate form ostendit ei stands italicized four lines later at 0233D in the same chunk. A dative pronoun with no change of sense, and Migne carries both — recorded, not marked.
0227D and 0229A · the two Alia littera readings (Dominus conteret bella; Misisti spiritum tuum et cooperuit eos mare). Migne has flagged both himself. Per the standing rule, no marker where he has already marked it. Both confirmed on the plate (f114 left, f115 left).

§6 · Readings examined at the plate and left alone

0221C · resurgant. Our twin reads resurgant; the Gallica impression at this point sets a heavily inked bowl that could be read o (resurgont), and a 14× crop did not settle it. The archive copy prints resurgant cleanly (l. 16978), our twin agrees, and the subjunctive is what the ut clause requires. No marker, no patch. Recorded so the Gallica image alone does not re-open it.
0229D · dicite … dicito. Migne addresses a singular throughout (talem cum videris … si quem videris) and then sets a plural imperative dicite the first time and a singular dicito the second. Both witnesses read dicite (Gallica f115 left, archive l. 17614) and our twin agrees. A real Latin form, so Pattern 7 governs and Pattern 12 has nothing to wrap; under Pattern 17 the plural is rendered as a plural and the singular as a singular, and the inconsistency stays visible on the page.
0231D · praeparatum. Quid est ergo hoc sanctimonium non manu hominum, sed Dei praeparatum. — an interrogative closed with a full stop. Migne's mark, kept (Pattern 8); this is not the 8b class, because English can carry the sentence as printed.
0233D · laudandus est. Sed in tali etiam ligno quis nisi creator et demonstrator laudandus est. Same class, same treatment.
0233C · concludit [condidit]. The square brackets are Migne's own, confirmed on the plate (f117 left column, corners 233/234). This is his inline conjecture layer, the [devicto eo] class at 0184B. The bracketed word rides into the English untranslated and unmarked; we do not stack a [cj:] or a [sic:] on his own mark.
0234A · the point after vocem Domini Dei. The Gallica impression looks like a full stop; the archive copy prints a comma and so does our twin. Not fired — and the same page is one where Gallica drops a hyphen (§ patches, scili cet), so its small marks on this leaf are suspect by the standing under-inking rule.
Foot-of-page apparatus: CHECKED ZERO on all seven leaves, both columns, both layers (numbered Forte notes and the asterisk/letter layer). What stands at the foot of f117 is the signature PATROL. CXIII. and the sheet number 8.

§7 · Register notes

ternistratores (Ex 15:4, Greek τριστάτας) is kept as italic Latin at all four occurrences (0228A ×2, 0228B ×2). It is a technical word with no English exponent — "third-riders", "chariot officers" and "chosen captains" all decide the etymology the gloss spends a paragraph deciding for itself — and the register rule for untranslatable technical Latin covers exactly this case.
The canticle is prose in Migne and is prose here. Ex 15 is set as continuous text on ff. 113–116 with no lineation of any kind; Pattern 8a forbids imposing verse lines, however plainly the Latin scans. Checked on every leaf.

Stint: CAPUT XVI–XIX — the manna, the water from the rock, Amalec, Jethro, and the approach to Sinai. All nine leaves of the range were read at the plate before anything below was written, and no column in the range had been read during prep.

§1 · Span count against the brief — 186 against 186, EXACT

data/briefs/8949-lemmata-0020.txt declares 186 marked spans for chunks 0020–0024. Counted in my own Latin: 186 whole-file, 186 paragraph-bounded. The agreement is not the false kind that cancelling errors produce, because this range contains 0 « and 0 » — there is no unmatched delimiter that could swallow a span in either direction. Per chunk: 0020 = 41 · 0021 = 36 · 0022 = 40 · 0023 = 38 · 0024 = 31.

Marker parity against the Latin twin, checked per chunk before each file was written: column anchors 14 / 10 / 13 / 12 / 10, identical strings in identical order; [n: …] 27 / 11 / 21 / 15 / 12, identical contents in identical order; one ## head per chunk, (cont.) untranslated.

§2 · Marker tally

marker	count	columns
`[var:]`	16	0235A · 0236A · 0237A · 0237D · 0238A ×2 · 0238C · 0239D · 0241C · 0241D · 0243A · 0243B · 0245A · 0246A · 0246B · 0248B · 0248C ×2 · 0249A
`[sic:]`	1	0246C
`[cj:]`	1	0245D
`[d:]`	0	—
`[ed:]`	0	—

Every column carrying a marker was read by me at the plate: 235, 236, 237, 238, 239, 241, 243, 245, 246, 248, 249. See data/briefs/8949-PLATE-READS-0020.json.

Six candidate [sic:] were withdrawn at the plate and proved to be our own digitization (§5); three further patches were refused because the witnesses split or the copies are a press variant (§6).

§3 · The 7a″ collation — ranked by whether the gloss is BUILT on the divergence

Tier 1 — the gloss expounds a word its own lemma does not print

0243A · Isa 65:2 (Rom 10:21). Migne prints Extendi manus ad populum non credentem and cites Isa. LXV. The Clementine's Is 65:2 reads Expandi manus meas tota die ad populum incredulum, and Rom 10:21, whose second half the printed words match exactly, reads Tota die expandi manus meas ad populum non credentem. Neither witness prints extendi, and the whole gloss turns on that verb: three words earlier Origen writes Moyses elevat manus, non extendit; Jesus … dicit: Extendi manus. The contrast lifting-vs-stretching is the exposition, and it rests on a verb the received text does not have. [var:] fired. ⚑ The 0015–0019 stint reports the same Isa. LXV address carrying Rom 10:21's wording at 0230B; this is the same site glossed a second time, with a third verb.

0246B · Eccl 1:8. Migne prints Non satiabitur oculus visu, nec auris auditu; the Clementine reads Non saturatur oculus visu, nec auris auditu impletur. The very next clause of the gloss is nec nos satiari possumus considerando quae scripta sunt — the argument's verb is taken from the printed form, and a rendering conformed to saturatur would leave the gloss echoing nothing. [var:] fired.

0248B–C · Ecl 9:8, 2Cor 7:1, Act 11:9 — three in one paragraph, all in the same direction. Origen is arguing that no one comes to the word of God but lotis vestibus, in clean garments, and Migne's three proof-texts have all been pulled toward mundus:

In omni tempore sint vestimenta tua munda against Ecl 9:8 candida, "white";
Purificatus es corpore, mundatus ab omni inquinamento against 2Cor 7:1 mundemus nos, a first-person exhortation turned into a second-person statement of fact;
Quae ergo Deus mundavit, tu immunda ne feceris against Act 11:9 tu ne commune dixeris, "do not call common" (Act 10:15 has Quod Deus purificavit). Three [var:]. Taken singly each is modest; taken together they are the argument, and the third is the strongest, because immunda ne feceris replaces a verb of speech with a verb of making in a passage about washing.

0239D · Ps 55:5. Migne prints Non timebo quid faciat caro; the Clementine has quid faciat mihi caro. Augustine's locution then quotes the SAME psalm two clauses later as quid faciat mihi homo — so the mihi Migne drops in the first quotation is retained in the second, on the same line, in a gloss whose whole point is the pairing of the two. [var:].

Tier 2 — Migne's address names one place and his printed words are another's

0241D · Zach 13:7 / Mt 26:31. Migne prints Unde Zach. VIII: Percutiam pastorem, et dispergentur oves gregis. Zach 13:7 reads percute pastorem, et dispergentur oves — an imperative, and no gregis. The printed words are Mt 26:31 verbatim. The gloss depends on the divergence in its own way: percussus enim Christus, et in cruce mactatus, Novi Testamenti fontes produxit needs the first-person future ("I will strike"), which is the evangelist's form, not the prophet's. [var:] naming the Zacharias reading. (The chapter number is Migne's own and rides as printed.)

0246A–B · Mt 12:42 / Lc 11:31. Migne prints Matth. XII: Regina quoque austri surget in judicio cum viris generationis hujus, et condemnabit eam. Mt 12:42 reads cum generatione ista … et condemnabit eam; cum viris generationis hujus is Luke's (Lc 11:31), which then closes condemnabit illos. So the printed sentence is a conflation that belongs to neither address whole. [var:]. This is the class the 0015–0019 stint names, and it is live twice more in my range in a form that takes no marker (§7).

0249A · Lc 8:10 / Is 6:9 — and it is a NEGATION. Migne prints ut videntes non videant, et audientes nos intelligant. Lc 8:10 reads et audientes non intelligant; the Clementine's Is 6:9, which is the address Migne gives, reads something else again (Audite audientes, et nolite intelligere). The printed nos is a real word, so Pattern 12 has nothing to wrap and the English renders it as printed: "and hearing they might understand us." Negation fidelity cuts the other way here — the Latin prints no non at that point, so the English carries none, and the [var:] is what tells the reader the received text has one. ⛔ See §6: the two copies of PL 113 differ at this word, and that is why no [sic:] was fired.

Tier 3 — real divergences, sense affected, gloss not built on them

0235A · 1Cor 10:9–11. Neque murmuraveritis, sicut quidam eorum murmuraverunt, et a serpentibus perierunt. Scripta sunt haec propter nos, in quos fines saeculorum devenerunt. v. 10 ends et perierunt ab exterminatore; a serpentibus perierunt is v. 9, about tempting, not murmuring; and v. 11 opens scripta sunt autem ad correptionem nostram. A three-verse conflation quoted as one. It matters because the gloss immediately threatens mittentur in eos serpentes.
0236A · Os 3:4. sine rege, sine principe, sine hostia, sine sacrificio, sine sacerdote against et sine altari, et sine ephod, et sine theraphim. Migne's list has a victim and a priest the prophet's has not, and drops three items the prophet's has.
0237A · Is 66:24. Vermis eorum non moritur under the address Isa. LXVI, where Isaiah has the future non morietur; the present is the form of Mc 9:43–47. ⚑ Worth noting that the same commentary then prints vermis qui non morietur at 0240B — Migne carries both tenses.
0237D · Ps 50:19. Cor contritum et humiliatum Deus non spernit against Deus, non despicies: a third-person statement where the psalm addresses God in the second person.
0237D · Mt 9:29. Fiat tibi secundum fidem tuam against the plural Secundum fidem vestram, fiat vobis. The gloss applies it to the single reader (fiat tibi quodcunque desideras), so the singular is doing work, but the singular is also the form of Mt 8:13 and 15:28 and may be memory.
0238A · Ps 36:8. derelinque indignationem against derelinque furorem.
0238A · Ps 112:7. Dominus allevat de terra inopem, et de stercore erigit pauperem against the participles Suscitans … erigens, and with a subject supplied that the psalm has not.
0238C · 1Jo 2:18. Pueri, novissima hora est against Filioli.
0241C · Ps 41:3. Sitivit anima mea ad te, Deus against ad Deum fortem, vivum.
0243B · Is 35:3. Remissas manus et genua debilia erigite under the address Isa. XXXV, where Isaiah reads Confortate manus dissolutas, et genua debilia roborate. The printed words are nearest Hbr 12:12 (remissas manus, et soluta genua erigite), and even there debilia is Isaiah's.
0245A · Gn 3:10. Audivi, Domine, vocem tuam, et abscondi me, quia eram nudus against Vocem tuam audivi in paradiso, et timui, eo quod nudus essem, et abscondi me — the fear and the paradise both gone, and a vocative Domine supplied.

§4 · The two markers that are claims about Migne

0246C · [sic: accepti] — and it is a [sic:], not a [cj:]. The plate prints sed Moyses, qui erat mansuetus super omnes homines (Num. XII), acce-|pti consilium inferioris, ut principibus daret formam humilitatis. Both witnesses agree: Gallica sets pti at the head of the next line, unmistakable at 10×, and the archive layer prints accepti. accepti is a real Latin form (gen. sg. / nom. pl. of acceptus) — but it has no grammatical slot here: strip it and the sentence from sed Moyses to daret has no finite verb at all. That is Pattern 12's first test for a real word, so the type is carried and, under the 2026-09-05 amendment, the sense is rendered beside it: received [sic: accepti] the counsel of an inferior. Conjecture: accepit. ⚑ This is only the second [sic:] in the book (the 0015–0019 stint has 0224D ververba), and it is fired against a page on which no other reading was doubtful and neither copy is worn.

0245D · [cj: indicabunt; read judicabunt, "shall judge"]. The plate prints Ordinabis, inquit, eos tribunos et centuriones, et indicabunt populum omni hora; at 4× the word opens with a clean in and no descender, and the archive layer prints indicabunt too — so it is Migne's, not ours. It is a real word that renders in place ("they shall point out the people"), so Pattern 12 has nothing to wrap. But the gloss two lines below is qui debent omni hora populum judicare, semper et sine intermissione sedere in judicio, and Ex 18:22 LXX is κρινοῦσιν. The printed word therefore makes the sentence say what Origen is arguing against, which is Pattern 18's bar; the gloss is additive and Migne's word keeps its English.

§5 · Six candidate `[sic:]` withdrawn — every one was ours

Filed in full at data/briefs/8949-PENDING-TEI-PATCHES-0020.md. In brief: ut surpra → ut supra (0237C) · languores tues → tuos (0238A) · postia → posita (0241A) · petrum ex qua → petram (0242A) · VERS. 12-25. → VERS. 12-15. (0244C) · crubescens → erubescens (0247A).

⚑ 0242A is the instructive one. petrum ex qua reads from our file alone as a gender disagreement of Migne's, which Pattern 9 exempts from marking — so the right decision (no marker) would have been reached for the wrong reason, and the Latin would have stayed permanently wrong. Only the plate distinguishes "Migne wrote a solecism" from "our transcription dropped a letter."

⚑ 0244C is the one nothing else could have caught. A verse address is not a word: no lexicon flags VERS. 12-25, and both scans' founts are ambiguous between 3 and 5 on this leaf. It was settled by the sequence — the next address in the same column is VERS. 16-18 — and confirmed by the archive layer, which distinguishes the two figures where Gallica does not.

§6 · Declined, and the reasons

0236A panis coelestis. qui and 0238A abscondit. vermes — Gallica shows a tailless point where our twin has a comma, on two consecutive leaves. The archive prints a comma at both. Two witnesses split; our twin sides with the archive; no patch, and no [sic:] either way. The finding that matters is that f118 and f119 under-ink comma tails, so every other call made on those two leaves was re-weighed.
0249A nos intelligant — the two copies of PL 113 are two printings here. Gallica prints non videant … nos intelligant; the archive, rendered as an image at 500 dpi, prints non vident … non intelligant. Two words differ on one line, both impressions clean, and our twin agrees with Gallica at both. There is no one to accuse: no [sic:], no patch. The English renders the reading our twin and Gallica share and a [var:] carries the received text. ⛔ This is the PL 114 "two printings, not two photographs" class appearing in PL 113, and it should be weighed before any future letter-level call in this tome is settled from one copy.
0243B non tam manum et armis — plate-confirmed. An accusative where the sense wants an ablative (manu), beside a correctly ablative armis. Declined: a case slip that renders in place ("not so much with hand and with arms") and asserts nothing false. Pattern 18's ruling explicitly keeps mood, voice, number and person anomalies out unless a negation or correlative is at stake.
0239B murmuranti et tentati Synagogae — plate-confirmed. tentati fails to agree with the dative Synagogae. Same class, same decline; English has no exponent for it.
0247C ac per hoc in ista umbra futuri secundum agni immolati diem festum; — plate-confirmed, a clause with no verb, closing on a semicolon. Rendered as closely as the printed words allow and not repaired. Nothing is missing from our twin; the ellipsis is Migne's.
0240C (AUG., ibid., quaest. 62. — the plate genuinely opens this attribution and never closes it. Same class as the book's one unclosed guillemet at 0192A, which the lemma brief forbids us to repair. Reproduced verbatim, untranslated; no marker, no patch.
0244A exonustare and 0241C sites (plural of sitis) — plate-confirmed, both attested medieval forms. Pattern 9: translated normally, no marker.
Migne's own citation slips, all plate-confirmed, all left to ride as printed under the refDisplay principle: psal. I at 0245A for Ps 50:6 (Tibi soli peccavi); psal. XIX at 0244A for Ps 9:7; Matth. XII at 0246A for Mt 5:22 (Reus erit gehennae ignis); (Matth. XII) at 0242D appended to a quotation the text itself introduces as Joan. XV; Isa. VI at 0249A for what is Lc 8:10's wording; (Matth. IV) at 0238A for what is nearest Ps 102:3. A wrong address beside right words is not a [var:] — Pattern 14 is about the quoted wording. Where both the address and the wording are in play the marker fired (0241D, 0246A–B, 0249A).
0238B–0239A Ascendent in montem, et bibent vinum, unguento ungentur. — plate-confirmed, and not traced to any Clementine verse. Recorded unidentified rather than forced onto Amos 6:6.
Band-letter arithmetic and repeated column numbers — not reported (runbook false positive 1). Every chunk boundary in the range was checked at the plate for text broken mid-word or mid-construction and all five are clean: 0019/0020 falls on a verse address at 0234D, 0024/0025 falls immediately before the CAPUT XX head at 0249C.

§7 · Two things about this range worth carrying forward

The fount's 3 reads as a 5 on the Gallica copy, and the archive copy is the corrective. On f122 the plate's VERS. 12, 13 renders as 12, 15 while its genuine VERS. 12-15 renders identically. Any numeral call in this tome needs the second witness, and the archive's OCR text layer — not just its image — distinguishes them.

Where the archive's OCR layer and its IMAGE disagree, render the image. At 0249A the OCR layer and the rendered page happened to agree, but the OCR layer had already manufactured a separate error on the same line elsewhere in this range (it drops malas from malas cogitationes at 0238B, which both our twin and Gallica carry). The layer is nearly free and worth grepping every time; it is never the last word.

Stint 0025–0029, 2026-09-08. CAPUT XX–XXII: the Decalogue itself, and the judgments that follow it. Every column on which a marker stands was opened at the plate (Gallica bpt6k5505319w, leaves f125–f131, printed corner numbers read first on all seven, both columns of every leaf). Plate reads are recorded in data/briefs/8949-PLATE-READS-0025.json; transcription and structural divergences in data/briefs/8949-PENDING-TEI-PATCHES-0025.md.

Span count. The brief declares 150 marked spans for this range. My own Latin counts 150 italic spans, and 150 again counted paragraph-bounded, with zero guillemets in all five chunks — consistent with the brief's statement that Exodus's single « stands at 0192A, outside my range, and with the paragraph-bounded check finding no unmatched delimiter anywhere. The totals agree, and they agree for the right reason.

Marker parity. Column anchors 12 / 14 / 1 / 15 / 6 and notes 14 / 24 / 1 / 24 / 11, each identical to the Latin twin and to the chunks' own noteCount.

1. ⭐ THE HEADLINE — the Decalogue's FIRST precept, and it is now marked

8949-ADJUDICATIONS.md §2 records the first precept as diverging from Exodus 20 and carrying no marker, because its only occurrence then in play stood at 0205B, inside a peer's chunk. It falls again inside my range, six times in chunk 0025, and it is now marked — once, at its first occurrence at 0250A, where Origen introduces it as Primum mandatum ergo est.

col	Migne prints	Ex 20	verdict
0250A	Non erunt tibi dii alii praeter me	20:3 Non habebis deos alienos coram me	`[var:]` fired here
0250B, 0250C ×3, 0251A, 0252B	the same, five more times	—	recorded, not re-marked

⭐ And Migne himself supplies the control, which is why this is a finding and not a suspicion. Eleven lines below the marker, at 0250C, his own verse lemma prints the Vulgate:

VERS. 3.-- Non habebis deos alienos, etc. [n: (ORIG.)] Primum mandatum est: Non erunt tibi dii alii praeter me;

The Vulgate wording and the Septuagint wording stand in one sentence, four words apart, on a column I read at the plate (f125, corners 249/250, right column, band C). So the divergence is not the compositor and not our transcription: Migne sets the Vulgate as the lemma and lets Origen expound the Old Latin/Septuagint form his argument actually needs — and it does need it, because the whole gloss turns on the tense:

Si dixisset, Non sunt, absolutior sermo videretur; sed quia dicit, Non erunt, non negavit quia sint, sed ne illi sint, cui haec praecepta dantur, inhibuit.

Non habebis has no future esse in it at all. Conform the lemma and the paragraph loses its subject. This is Pattern 7a″'s founding shape, and it is why the marker sits on the gloss's wording rather than on Migne's lemma.

⚑ The marker fires ONCE. Six markers on six occurrences of one three-word phrase inside 1,400 words would be apparatus, not annotation. The note at 0250A says the wording recurs; the crux says where.

2. The rest of the `[var:]` markers — twelve in all, with the bar I applied

Every one stands on a column I opened. Where a class recurs (Origen's Old Latin prophets, patristic paraphrase) I state the class once here and fire nothing, following the 0010–0014 stint's precedent.

The bar, stated so it can be checked: fire where the received text at the address Migne gives says something different in kind, or where the gloss's own argument stands on the printed word. Decline where the divergence is a synonym, a spelling, a word order, or a substituted noun inside a list that no line of the exposition touches.

2a. The Decalogue and its lemmata

0250C — Ex 20:4, and the exposition is built on the divergent noun. Migne: Non facies tibi idolum, neque ullam similitudinem. Ex 20:4: Non facies tibi sculptile, neque omnem similitudinem. Origen then spends a full column separating the two nouns — Aliud est ergo facere idolum, aliud similitudinem, and idolum is the word he defines (facit enim quod non est). Restore sculptile and the distinction has nothing to stand on. Plate-confirmed f125 c2 band C, and again at 0251B in Origen's fuller re-quotation.
0255A — Ex 20:17 quoted in Deuteronomy's words, inside a commentary on Exodus. Migne: Non concupisces uxorem proximi tui? The pre-scan ticks this ✓ Dt 5:21, and the tick is the finding: Ex 20:17 reads Non concupisces domum proximi tui, nec desiderabis uxorem ejus. Same class as the 0010–0014 stint's 0215D. Plate-confirmed f128 c1 band A.
⚑ 0255B — the internal control that vindicates the peer's Non furaberis finding. Migne here prints Non furtum facies, Ex 20:15 exactly, plate-confirmed on the same column (f128 c1 band B). So Rabanus's Non furaberis at 0211A is not a compositor's slip: the same book, eleven columns later, sets the Vulgate form correctly when Augustine is quoting. That is positive evidence for the 0010–0014 stint's reading of the class as a catechetical short form. No marker; recorded as a control.
⚑ 0255A — the œ ligature again, and no false claim was made. The plate prints Non mœchaberis four times on that column (also mœchandi, mœchetur), exactly as ADJUDICATIONS §3 warns. Our twin has oe here, so nothing was ever at risk — but the check was run before any Decalogue verb was touched, and it is recorded so the class stays closed.

2b. Proof-texts whose argument stands on Migne's word

0250D — Dt 32:8, and this is the most consequential divergence in the range. Migne: statuit terminos gentium secundum numerum angelorum Dei. Dt 32:8: juxta numerum filiorum Israel. The Septuagint reading, and the entire following paragraph is nothing else: Angelos ergo, quibus gentes commisit Excelsus, dominos vel deos appellat Scriptura. With filiorum Israel there are no angels in the verse and no argument on the page. ⚑ Corroborated inside the volume itself: the archive OCR text layer carries a second, unrelated occurrence elsewhere in PL 113 — Statuit terminos gentium juxta numerum angelorum Dei (Deut. xxxii) — so angelorum Dei is the form this tradition transmits, not a local slip. Plate f125 c2 D.
0250D — Ps 49:1, where the singular destroys the proof. Migne: et Deus Deo locutus est. Ps 49:1: Deus deorum Dominus locutus est. The verse is adduced, with Rex magnus super omnes deos, to show deos nominari — Scripture naming gods in the plural. Deus Deo names none. Plate f125 c2 D.
0254A — Ps 88:34, and the plural is what carries the argument. Migne: misericordiam autem meam non auferam ab eis. Ps 88:34: non dispergam ab eo. Origen's point is that God visits the sins of the fathers on the sons ut, purgati, diabolo in poena socii non efficiantur — the mercy has to reach them, the sons. Plate f127 c2 A.
0252A — 3 Kg 18:21. Migne: Usquequo claudicatis utroque pede? against in duas partes. A different figure, and the sentence after it works the figure (Ab illis ergo quasi fugaces caedimur). Plate f126 c2 A.
0252A — Jer 4:30, where the received verse says something else altogether. Migne: unde Jer. IV: Amatores tui inimici facti sunt tibi. Jer 4:30: contempserunt te amatores tui; animam tuam quaerent. The sentence Migne prints stands nearest Lam 1:2 (facti sunt ei inimici). Plate f126 c2 A.
0256C — 1 Pet 2:5, indicative for imperative and houses for a house. Migne: Vos estis lapides vivi, coaedificati in domos spirituales. 1 Pt 2:5: et ipsi tamquam lapides vivi superaedificamini, domus spiritualis. In a gloss whose entire subject is that the unhewn stones in unitate solidantur and omnes unum altare faciunt, the plural is worth showing. Plate f128 c2 C.
⭐ 0258A — Origen's Ex 21:22–24, and the Vulgate does not contain the distinction the gloss expounds. Migne prints the Septuagint text: …et exierit infans ejus nondum formatus, detrimentum patietur… Quod si formatus fuerit, dabit animam pro anima, oculum pro oculo. The Vulgate has abortivum quidem fecerit, sed ipsa vixerit… Sin autem mors ejus fuerit subsecuta — no formed/unformed distinction at all. Both the Augustine quaestio immediately above (Quod non formatum puerperium noluit ad homicidium pertinere) and the two columns of Origen below it are about nothing else. Plate f129 c2 A.
⭐ 0259B — 1 Cor 3:15, and the argument uses Migne's word five times. Migne: Si cujus opus arserit damnum patietur. 1 Cor 3:15: detrimentum patietur. Around it: Damnum ergo patietur, qui excussit. De damnis ergo doctorem dicit Apostolus … quaedam peccata ad damnum pertinent, non tamen ad interitum … qui damnum passus fuerit … damno dicitur percuti. Plate f130 c1 B.
0259B — Mt 16:26, and it is the pair to the last, not a second slip. Migne: animam autem suam perdat where Mt 16:26 has animae vero suae detrimentum patiatur. ⚑ Both quotations in this one paragraph are adjusted AWAY from detrimentum — one toward damnum, one toward perdat — which is what makes them a matched pair serving the distinction the gloss then draws, ad damnum … non tamen ad interitum. Plate f130 c1 B.
0260B — Mt 24, two verses welded into one sentence. Migne: Vae praegnantibus et nutrientibus in illis diebus, in quibus scandalizabuntur, si fieri potest, etiam electi. Mt 24:19 ends at in illis diebus; the remainder is Mt 24:24 (ita ut in errorem inducantur (si fieri potest) etiam electi), joined by a relative neither verse has. Plate f130 c2 B.
0260C — Prov 16:23, hominis for sapientis. The proverb's whole force is that it is the wise man's heart. Plate f130 c2 C. (See §5 for the separate, unresolved question about the verb in the same clause.)
0261D — Ex 22:25, and the lemma will not construe. Migne: Si pecunia mutuam, etc. — nominative with accusative. Ex 22:25: Si pecuniam mutuam dederis. ⚑ Corroborated independently by the archive.org OCR text layer, which reads Si pecunia mutuamy etc., so both witnesses agree the m is absent and this is not one photograph's ink. Plate f131 c1 D. Marked [var:], not [sic:]: the type is legible and both words are real Latin words, so Pattern 12 has nothing to wrap; Pattern 14's test — legible, complete, and not agreeing with the Vulgate — is met exactly.

3. ⛔ ZERO `[sic:]`, and that is the finding

No [sic:], no [d:], no [ed:], no [cj:] fired anywhere in this range. This book has now produced none across three stints, and every candidate I met went the same way the 0005–0009 stint's four did: it looked like broken type from our twin and was Migne's own, or could not be settled on one witness. The candidates, with their disposition:

0254A in futurum. unde Matth. VIII): — a full stop, a lowercase unde, and an unmatched closing parenthesis, all in nine words. From the twin alone this is three digitization failures in a row and reads exactly like the 0005–0009 stint's apud. Deum. The plate sets all three (f127 c2 band A). Migne's. Pattern 8 preserves them; the English keeps the stop and the lowercase whence. No marker, no patch.
0261D Si pecunia mutuam — handled as [var:] above, not [sic:], for the reason given.
0260C crudiet — see §5; deliberately unresolved.
0249D qui eduxi te terra Aegypti, with no de, against de terra in the full form at 0250A. Plate-confirmed as Migne's (f125 c1 D). A bare ablative of separation is sound late Latin, the English is identical either way, and Migne is abridging (etc. follows). No marker.
0249D domus liberalitatis, standing between three occurrences of domus libertatis. Plate-confirmed as Migne's (f125 c1 D, re-cropped at 3.4×). Liberalitas is a real word, so Pattern 12 has nothing to wrap and Pattern 18's census is closed; the neighbouring libertatis makes the slip visible to any reader. Rendered literally, "the house of liberality."
0256B opera manum hominum for Ps 113:12's manuum. Plate-confirmed (f128 c2 B). One letter, both forms real, no sense change, nothing in the gloss touches it. Declined.
0251B nil est idolum beside Migne's own nihil est idolum nine lines later at 0251C. Both plate-confirmed. Pattern 9 — an attested contraction, not a defect. Declined.
0255C Cunc'us for Cunctus on the plate — a worn t, and our twin reads Cunctus correctly. Under-inking, not type. Declined.

4. Migne's own oddities — recorded, untouched, and none of them ours

Each is plate-confirmed and each would read as a build defect to anyone who did not open the page.

0257A VERS. 1. 6.-- — a period between the two verse numerals, where every neighbouring head prints a comma (VERS. 16, 17.) or a range dash (VERS. 22-28.). Migne's (f129 c1 A).
0260A [n: (Psl. III)] — with (Psal. LVII) printing correctly two lines below on the same column. Migne's (f130 c2 A). [n:] contents ride verbatim on both sides; no patch.
0261C — the Rabanus gloss on Ex 22:18 ends …donec maleficium erroris in eis moriatur with no terminal full stop. Migne's (f131 c1 C). Pattern 8 forbids supplying one, and the English ends without a stop.
0262A (AUG., quaest, 86 in Exod.) — a comma after quaest, against (AUG., loc. 99.) six lines above. Migne's (f131 c2 A). Same class as the 0010–0014 stint's (ORIG.) hom. 4 in Exod.). What is ours at that site is that the TEI never marked the parenthetical as a note at all; see the patches file.
Mis-bound [n:] citations, all his. 0251B (Joan. V) on a quotation of Jn 8:34; 0254D (Ezech. XXXVI) on wording that agrees with Ez 11:19's dabo eis rather than Ez 36:26's dabo vobis; 0258C ibidem covering a conflation of 2 Tim 2:23 with Tit 3:9. Recorded only.

5. ⚠ Two readings left open, and neither was decided on one witness

0260C — crudiet or erudiet. At 6× native the Gallica plate reads Cor hominis crudiet os ejus: c-r-u-d-i-e-t, with no e before the r, in the same italic as Cor immediately preceding. If that is the type, it is a non-word and the first [sic:] this book would have produced. It is not fired. The archive.org OCR text layer reads erudiet, as does our TEI; the archive copy is 1-bit JBIG2 symbol-substitution, whose errors are precisely well-formed words that construe, so it cannot settle a letter — but an e losing its crossbar to read as c is also the classic ink-loss shape, and a [sic:] is a public accusation against Migne founded here on one photograph. The 0010–0014 stint's disposition of the Greek at 0219B is the precedent and I have followed it exactly: no marker, no patch, the English carries the twin's reading ("shall instruct"), and the site is filed for a second physical witness. ⚑ The separate [var:] on the same clause (hominis for sapientis) is uncontested by both witnesses and stands on its own.
0250C — VERS 3. with no period after VERS. Our twin prints VERS. 3.--. If the plate is right, the patch runs the opposite way from the usual one — removing our period, not adding his — because the 0005–0009 stint established that VERS 20. at 0201C, missing its period, is Migne's. But it is a single fine point on a book proved under-inked elsewhere, and that stint's own rule is that a page under-inked at one site is suspect at every site on it. Not acted on in either direction.

6. ⭐⭐ A FINDING ABOUT THE BOOK, NOT ABOUT MY RANGE — Migne has a foot-of-page note layer here

Full detail in data/briefs/8949-PENDING-TEI-PATCHES-0025.md; the substance in three lines, because it changes what a later pass should assume.

At 0256C the plate keys a raised letter ª to the last words of the Isidore gloss, and the foot of leaf f128 carries a long two-column editorial note — running over onto the foot of f129 — on the division of the Decalogue: Augustine's enumeration, the scholastic doctors who follow it, and Origen's conjoining of the two prohibitions of concupiscence. Our TEI carries neither the key nor a word of the note.

It is Pattern 19's letter-keyed layer (the 8950 Genesis shape), and it is prose, not a reading, so it takes a Latin [cn: a | …] and a translated English [nt: …].
⛔ It overturns an inherited zero. CLAUDE.md records PL 113's "no foot-of-page apparatus" finding as resting on 19 pages across four books, with the standing warning never inherit a sibling book's zero. 8949 has the layer. Six of my seven leaves carry nothing at the foot, so it is sparse, not absent — which is exactly the distribution a spot check misses.
⛔ Nothing was recovered or invented. Recovery is per volume at chunking; Migne's raised key is not carried into the English in any case, so 0027.md is correct as shipped and stays correct after recovery.
⚑ And the note answers the question my own columns are arguing about. 0250B poses Augustine's quomodo decem praecepta legis dividenda sint, 0254C gives Isidore's division, 0255A–B gives Augustine's, and 0250C gives Origen's — and the note at the foot of the same opening adjudicates among them. A translator working from our files alone cannot see it. That is the 11613 F. earum situation exactly, one apparatus layer over.

7. Renderings decided, and why

⭐ praeceptum = "precept", mandatum = "commandment", held apart throughout — the provisional convention of ADJUDICATIONS §1, and my range gives it a reason the 0010–0014 stint did not have. See §8.
colere = "to worship", adorare = "to adore", never interchanged. At 0252C the gloss exists only to separate them: Aliud est colere, aliud adorare … Utrumque ergo resecat, ut nec affectu colas, nec specie adores. Collapsing them costs the paragraph its subject.
zelans = "jealous", zelus = "jealousy", zelotypia = "jealous rivalry". The 0252D–0253B passage is a sustained marriage figure and uses all three within a few lines (nec zelotypia movetur … excitatur super eam zelus ejus … Deus zelans). Three Latin words, three English ones; zelus is not rendered "zeal" anywhere here, because in this passage it is the husband's jealousy, not eagerness.
sermo in the Decalogue frame = "word" / "saying", not "sermon" or "discourse." Isidore says the law in decem verbis consistit, so primus sermo decalogi (0249D) is "the first word of the decalogue" and non est sermo mandati (0250A) "not the word of a commandment". At 0250C absolutior sermo videretur is "the statement would seem more absolute" — the same word doing ordinary work, and rendered ordinarily.
damnum = "loss" throughout 0259A–C, never varied to "detriment" or "harm", because the whole exposition is a definition of the word (see §2b). Detrimentum appears only inside the [var:] notes, which is where the Vulgate's word belongs.
similitudo = "likeness", idolum = "idol", sculptile only inside a [var:]. Same reasoning as damnum: Origen is defining the terms.
subreptio (0254C) = "stealthy deceit." Isidore's ten-item list pairs each commandment with the vice it kills, and the first commandment is against false gods; the word carries the sense of a thing got in by stealth rather than "theft" (his seventh item is rapacitas, his eighth falsitas, so the vocabulary is already crowded and each has to stay distinct).
Pattern 17 applied throughout. thou/thee/thy for every singular address — 0249D–0252B is a long second-person-singular exhortation (audire non poteris, si tibi injungatur, mereberis, Tibi ergo, o Israel, quibus in baptismo renuntiasti, Nisi ergo tota fide permaneas cum viro tuo), and you only where the Latin is plural (0249C veritas liberabit vos, 0252B Isa 52:11 exite … et ego recipiam vos, 0253C Vos de patre diabolo estis).
0252C anchor placement. Our twin sets [0252C] immediately before VERS. 5-11. with no space; reproduced exactly so, in place.
0257A — our twin sets a bare [0257A] on its own line between the head and the first lemma; reproduced as its own line, since moving it into the paragraph would change anchor position.
0262A — the italic run (AUG., quaest, [0262B] 86 in Exod.) is carried verbatim and untranslated, anchor inside, because it is one of Migne's citations that the TEI failed to mark as a note (patches file). Translating it would have made our page assert a note structure the Latin twin does not have.
Question marks are exact against the twin in all five chunks (0025 3/3, 0026 4/4, 0027 0/0, 0028 5/5, 0029 0/0). None supplied, none deleted.

8. ⭐ FOR THE MERGE — new evidence on praeceptum / mandatum, and it cuts one way

ADJUDICATIONS §1 leaves this open, with the 0010–0014 stint's argument (mandatum stands eleven lines from praeceptum and needs a different English) against chunks 0002 and 0009, which render both "commandment". My range settles the empirical half of it, and the answer is not close.

Migne uses both words for the Decalogue, in the same glosses, and the two enumerated series use different words.

Origen's series is mandatum. 0250A–0250C: Primum mandatum ergo est … Est ergo primum mandatum … Secundum … a primo mandato … non complebitur numerus decem mandatorum. Nine mandat- forms against three praecept- in chunk 0025.
Rabanus's series, in the peer's chunks 0010–0014, is praeceptum — secundum, tertium, quartum … praeceptum, fourteen of them.
⭐ And Isidore uses BOTH, four lines apart, at 0254C — plate-confirmed on f127 c2 band C:

Primum mandatum prohibet subreptionem; secundum, errorem; … decimum, adulterii cogitationem. Et notandum quod sicut decem plagis percutiuntur Aegyptii, sic decem praeceptis scribuntur tabulae … vel quia sunt praecepta duo dilectionis Dei et proximi … In una enim tria praecepta, quae ad dilectionem Dei pertinent.

Under "commandment for both," that paragraph reads as one word repeated seven times and the reader cannot see that Migne wrote two. Under the split it reads as what it is: the ten mandata enumerated, then the two tables written with ten praecepta, then the two praecepta of love. The distinction is doing work on the page, not in a translator's ear.

A second consequence, and it is the one that matters for the merged book. Because Origen's enumerated series is mandatum and Rabanus's is praeceptum, the split convention keeps the two series apart in English — a reader meeting "the first commandment" in CAPUT XX and "the second precept" in CAPUT VIII can tell that these are two different expositions of the Decalogue by two different authors. Collapse them and the book appears to contain one enumeration that keeps restarting. That is a stronger argument than the eleven-lines one, and it is available only once both ranges are in view — which is what the merge is for.

Counts, so the sweep can be sized: 28 mandat- and 49 praecept- forms across the 44 Latin chunks. I have applied the split uniformly in 0025–0029 (11 "commandment", 10 "precept", each tracking its Latin exactly). Chunks 0002 and 0009 are the known non-conforming sites.

▶ Recommendation to the merge: adopt praeceptum = precept, mandatum = commandment, sweep all 44 chunks, and record it in the conventions header. I have not edited a peer's chunk.

9. Declined as a class, once, so it is not re-raised

Origen's and Augustine's scripture is quoted from their own Old Latin throughout this range, and marking each occurrence would put a [var:] on every patristic quotation in the book — the 0010–0014 stint's ruling, which I have followed. The sites, all plate-confirmed and all declined: 0249C Jn 8:31–32 (abridged, permanseritis/agnoscetis for manseritis/cognoscetis); 0250D 1 Cor 8:5 (Si quidem sunt for Nam etsi sunt, dicuntur for dicantur); 0252A Sir 2:14 (duplici animo for duplici corde); 0252B Hos 2:5 (vinum meum for panes… et aquas meas); 0252B Isa 52:11 conflated with 2 Cor 6:17 (et ego recipiam vos); 0253A Isa 50:1, Jer 3:1; 0253B Jer 6:8; 0253C Jn 8:44 (de patre for ex patre), 1 Jn 3:9; 0254A Ez 33:11 (the current liturgical short form); 0254B 1 Cor 11:32 (Cum castigamur for Dum judicamur, uncited); 0254C Mt 22:40 (in quibus tota lex — Migne's adaptation to his own relative clause); 0258C Mt 7:3 (a statement where the Vulgate has a question), 2 Tim 2:14 (plural Nolite for singular Noli — and the plural is the adaptation to the two litigants, which is the point); 0258D Isa 26:18; 0259A Isa 66:8, Isa 37:3, Gal 4:19 (word order); 0260B Mt 18:6; 0261A Isa 53:7 (ductus est, which is Acts 8:32's form); 0261B 2 Cor 5:11, 2 Cor 12:19; 0261C Mt 5:26 (third person for second — and the gloss's subject is the shepherd, so the adaptation is deliberate).

Also declined, and for the reason the runbook's false-positive list gives: the band gap at 0257B → 0257D in chunk 0028. The A–D marks are positional, not four per column; no text is broken mid-word or mid-construction across it. No marker, no patch, no plate query.

Stint: CAPUT XXIII (from VERS. 3) through CAPUT XXVI — the close of the covenant code, the ascent of Sinai, and the tabernacle: the ark, the propitiatory and cherubim, the table and its loaves, the candlestick, the curtains, the tent-cloths, the boards and their bases, the veil and the hanging.

All eight leaves of the range were read at the plate on BOTH witnesses, both columns, printed corner numbers first, before anything below was written — Gallica bpt6k5505319w f131–f138 (261/262 … 275/276) and the archive copy patrologiaecurs04migngoog pp. 136–143 (same columns, corners likewise read first). Only 0265A–0266D had been read during prep; every other column here was unread. Detail per leaf: data/briefs/8949-PLATE-READS-0030.json.

§1 · Span count against the brief — 266 against 266, EXACT

data/briefs/8949-lemmata-0030.txt declares 266 marked spans for chunks 0030–0034 — "the densest range in the book". Counted in my own Latin: 266 line-bounded, 266 paragraph-bounded. The agreement is not the false kind the brief warns of: this range contains 0 « and 0 » (consistent with the book's single guillemet standing alone at 0192A), so there is no unmatched delimiter that could let two counting errors cancel. Per chunk: 0030 = 27 · 0031 = 36 · 0032 = 93 · 0033 = 68 · 0034 = 42.

§2 · Marker tally

marker	count	columns
`[var:]`	10	0262D · 0265A · 0267D · 0268A ×2 · 0269A · 0269B · 0270B · 0273B · 0274D
`[sic:]`	0	— (one candidate fired and withdrawn at the plate, 0262D)
`[cj:]`	0	— (two candidates declined, 0263A and 0272C)
`[d:]`	0	— (one candidate declined, 0270D)
`[ed:]`	0	—

Every column carrying a [var:] was read at the plate on both copies.

⭐ Five pending TEI patches, and all five are OURS (data/briefs/8949-PENDING-TEI-PATCHES-0030.md): a mos → annos at 0262D, Tabarnac. → Tabernac. at 0265D, and three Hebrew words Migne prints that our source drops — תשרי at 0263B, כרוב and כרובים at 0267D.

§3 · ⭐⭐ THE HEBREW LAYER OF THIS BOOK IS NOT COMPLETE — three more words, one of them load-bearing

The lemma brief states as settled that Corpus Corporum "drops every piece of Hebrew Migne sets inline" and that "all of it was read at the plate on 2026-09-08 and restored before chunking": the book-title banner plus six words at 0183B, 0193A, 0219A, 0266A (two) and 0296A. That inventory is short by at least three.

0263B — quem Hebraei Chaldaico vocabulo thisri, תשרי Macedones ὑπερβεραῖον vocant.
0267D — Numero singulari cherub, כרוב plurali autem cherubim כרובים dicitur.

All three are on both copies. The two at 0267D are unpointed, exactly as the brief describes Migne's practice in this book; תשרי appears to carry at least one sub-letter point on both copies and neither resolves it at maximum available resolution, so the consonantal form is what is carried and the vocalisation is flagged open in the patches file rather than guessed. Migne's own transliterations — thisri, cherub, cherubim — stand beside each and are the on-plate control; all three agree with the Masoretic forms.

⭐ The 0267D pair is not decoration, it is the gloss's evidence. Bede's whole point there is the singular and plural forms of the Hebrew word: Numero singulari cherub, כרוב plurali autem cherubim כרובים dicitur. With the Hebrew dropped, the sentence tells an English reader only that one Latin form is singular and the other plural — which he could already see — and the observation becomes trivial. This is the same shape as a [sic:] swallowing the force of what it quarantines (7a⁗), arriving by a different road.

⚑ And the class is cheaply greppable, which is the finding that generalises. Corpus Corporum drops the Hebrew but leaves the whitespace: all three sites show in the XML as a run of two or three spaces immediately after the Latin transliteration (thisri, Macedones, cherub,</hi> plurali, cherubim</hi> dicitur). A whitespace-run scan over the whole of 8949 should be run before this work ships; I found these by reading, and reading will not find the ones outside my eight leaves.

§4 · ⛔ ZERO `[sic:]` — and the reason is that the ARCHIVE COPY IS THE WORSE WITNESS ON THIS QUIRE

The standing guidance (PL113-WITNESS-ENCODING.md) is that the archive copy is 1-bit JBIG2, a symbol-substitution codec, and may never be a sole witness for a letter, digit or point. Measured here, over eleven sites where the two copies disagreed with each other or with our twin:

column	archive copy	Gallica and our twin	what firing on the archive would have produced
0263B	`seculo curantes`	`sedulo curantes`	a `[sic:]` on a phrase that does not construe
0263B	`amplus tuba canunt`	`amplius tuba canunt`	a `[sic:]`
0263C	`Ecce ergo mittam angelum`	`Ecce ego mittam angelum`	a false `[var:]` against Ex 23:20
0263D	`invidita`	`invidia`	a `[sic:]` on a non-word
0268A	`labium , etc.`	`labium., etc.`	a patch deleting pointing Migne prints
0269D	`Omnes opus candelabri`	`Omne opus candelabri`	a `[sic:]` on a number disagreement
0271B	`Fecit et quinquaginta circulos`	`Facies et quinquaginta circulos`	a false `[var:]` against Ex 26:6
0272A	`VERS. 6.`	`VERS. 9.`	a false claim about Migne's own numbering
0274B	`(Marc. XLVI)`	`(Marc. XVI)`	a `[sic:]` on an impossible chapter number
0262D	`annos` (clean)	Gallica: `a nnos` (loose sort)	— archive right, our twin wrong
0272C	`ad altera`	`ad altera`	— the two copies agree; Migne's

Nine to two against the archive copy. Every one of its nine errors is a well-formed word or number that construes — Ecce ergo mittam angelum reads perfectly, and it is not what Migne set. This is why nothing on this quire may stand on that copy alone, in either direction, and it is also the reason the range ships zero [sic:]: every candidate that survived our twin died at Gallica.

⚑ The reciprocal, and it matters: at 0262D the archive copy is the better witness. Our twin prints Tribus vicibus per singulos a mos; the archive prints annos cleanly; Gallica shows the same five letters (four distinct strokes after the gap, read at 18×, nn and not m) with an abnormally wide space after the initial a — a loose sort in that impression. The letters agree across both copies, so Migne's text is annos, our twin is wrong about the space and the letters, and the [sic:] was withdrawn. Migne's own correct annos prints six lines below in the second lemma of the same verse. "Our copy is poor" is a claim about a page, never about a scan.

§5 · The `[var:]` list — ranked by how much of the gloss stands on it

Ten markers. The first four are the ones the exposition is built on; the rest are wording.

⭐ 1. 0267D — Hab 3:2 · In medio duorum animalium innotesces

Migne prints, and both copies confirm: unde Habacuc ait: In medio duorum animalium innotesces (Habac. III). The Vulgate reads in medio annorum notum facies, "in the midst of the years thou shalt make it known". The Old Latin's two living creatures is the entire reason the verse is here: Bede is expounding the two cherubim standing over the propitiatory, and the proof-text works only in the form Migne prints. With the Vulgate wording the citation is a non sequitur. [var: Vulg. Hab 3:2 in medio annorum notum facies…] — the most valuable divergence in the range.

⭐ 2. 0270B — Ex 26:1 · Tabernaculum vero ita et fiet

The opening lemma of CAPUT XXVI, plate-confirmed. The Vulgate has Tabernaculum vero ita facies, "the tabernacle thou shalt make thus" — God's second-person command to Moses. Migne's impersonal passive turns the command into a description of how the tabernacle will come to be, at the head of the chapter whose whole burden is what Moses was ordered to build.

⭐ 3. 0269B — Ct 6:10 · Descendit in hortum nucum

Vulgate: Descendi in hortum nucum, "I went down into the garden of nuts", spoken by the Bride. Migne's third person moves the words out of her mouth into a narrator's. The gloss that follows reads the nut as the Church whose sweetness is hidden under a shell — an identification that in the Vulgate the Church herself makes about herself.

⭐ 4. 0265A — Nm 11:16 · Assume tibi septuaginta viros

Plate-confirmed. Vulgate: Congrega mihi septuaginta viros de senibus Israel, "Gather unto me seventy men of the ancients of Israel". Migne's Assume tibi — "take unto thee" — is a different verb and a different beneficiary, in the one place the gloss appeals to the literal sense (Vel ad litteram) against its own numerological reading of the seventy.

5. 0273B — the citation names Matth. XXI and the words are Mt 26:56

The wrong-verse class the common brief names, and the second instance found in this book. Migne prints Hoc autem totum factum est, ut adimplerentur scripturae prophetarum and keys it (Matth. XXI). Those words are Mt 26:56; Mt 21:4 reads Hoc autem totum factum est, ut adimpleretur quod dictum est per prophetam dicentem — near enough to be the source of the confusion, and different in exactly the word (scripturae prophetarum) the gloss is using to prove that the prophets are the bases of the apostolic boards. Both copies print XXI.

6. 0268A — Ct 2:4 · Introduxit me rex in cellam vinariam, et ordinavit in me charitatem

Plate-confirmed on both copies. The Vulgate has neither rex nor the connecting et: Introduxit me in cellam vinariam; ordinavit in me caritatem. The added rex is not idle in a gloss on the table of the tabernacle read as Christ's banquet.

7. 0268A — `VERS. 23.` printed twice, the second over Ex 25:24

Migne sets VERS. 23.-- In altitudine cubitum (correctly Ex 25:23) and then, three lines later, VERS. 23.-- Faciesque illi labium., — which is Ex 25:24 — after which VERS. 24. stands over Ex 25:25. His numbering lags by one for two entries and then rejoins. Both copies print 23 twice, so it is his, not a digitization slip, and it is left exactly as printed with the divergence noted beside it.

8. 0269A — 2Cor 13:13 · sit semper cum omnibus vobis

Migne adds semper, which the Vulgate has not (sit cum omnibus vobis. Amen.). Plate-confirmed. Minor, but it is a benediction quoted as a benediction.

9. 0262D — 2Cor 9:6 · in benedictionibus, in benedictionibus et metet

Migne repeats the preposition of the first member where the Vulgate turns it: qui seminat in benedictionibus, de benedictionibus et metet. Plate-confirmed on Gallica at 3.2×.

10. 0274D — Ex 26:24 · Eruntque sibi conjunctae

Vulgate: Eruntque conjunctae a deorsum usque sursum. Migne's sibi makes the joining mutual — which is precisely what the gloss then argues (the boards' framing would recede ab invicem if one prophet denied what another said), so the divergence and the exposition move together.

Checked and DECLINED — divergences that do not change the sense

0266A non venit legem solvere, sed adimplere (Mt 5:17). Migne adds legem and puts the verb in the third person, but the clause is woven into the author's own sentence (quia Christus…), and legem is supplied from Mt 5:17's own first half. Not a divergence, an adaptation.
0266A Si crederetis Moysi, forsitan crederetis et mihi (Jn 5:46) — word order of forsitan only.
0270A Si circumcidimini for Gal 5:2's circumcidamini; indicative for subjunctive, sense unchanged.
0269D dextera ejus for Ct 2:6's dextera illius.
0263C Initium omnis peccati superbia — Sir 10:15 has est between; ellipsis, not divergence.
0273A Ex 26:16 in latitudine cubitum ac semissem against the Vulgate's singulos ac semissem; different words, identical measure, and the gloss (which counts the ten cubits of length) does not touch it.
0264B Alia littera: Benedicam panem et vinum tuum and 0264C Alia littera. Emittam vespas ante te, et ejiciet Amorrhaeos — ⚑ Migne labels both divergences himself. No marker is stacked on his own mark (the standing rule from stint 0005–0009).
0272D De ianthinis pellibus for the Vulgate's janthinis — orthography, Pattern 9.
0266C Ex 25:3 quae accipere debetis for debeatis — indicative for subjunctive.

§6 · ⚑ THE TABERNACLE VOCABULARY — fixed here, and chunks 0035–0039 should follow it

Ex 25–26 is dense in technical terms for the structure and its furniture, and the 11551 precedent (one agent anglicising a column's members to "epistyle"/"shaft" while the other half of the work kept stylus/epistylium, breaking the author's own etymological argument) is the failure this table exists to prevent. Where a gloss's argument runs through a word, the rendering that keeps the argument intact beat the more idiomatic one. Every term below is fixed once and used the same way in all five chunks.

Latin	English	why this one
tabernaculum	tabernacle
arca	ark
⭐ propitiatorium	propitiatory	0267C is an etymology: sic dicta quia ibi Deus… propitiabatur populo. "Mercy seat" is the idiomatic English and it destroys the gloss, which explains the name from propitiari. Non-negotiable.
⭐ cherub / cherubim	cherub / cherubim	0267D is a gloss on the singular and plural forms; both must stay recognisably the same word, and Migne's own Hebrew (now recovered) stands beside them.
mensa	table
⭐ panes propositionis	loaves of proposition	0268D turns on it: bene panes propositionis vocantur, quia semper debet patere sermo salutis… quod Dominus… mundo proponens. "Shewbread" severs the link to proponere; "loaves of proposition"/"proposing" keeps it.
labium (of the table)	lip	literal; kept distinct from corona, which stands beside it in Ex 25:24–25
corona	crown
circulus	ring
vectis	bar
lamina	plate
acetabulum	dish
⭐ phiala / cyathus	bowl / cup	0268B–C is Gregory's contrast between the teacher who quasi phialam porrigit and the one who per cyathum gustum praebet: the whole point is that one vessel is large and one small. Two distinct English words are required, and "bowl" must be the bigger.
⭐ scyphus	goblet	Douay renders scyphi "bowls", which in this chunk would collide head-on with phiala. "Goblet" keeps the candlestick's cups apart from the table's vessels three paragraphs away.
sphaerula	little sphere	literal (a diminutive of sphaera); "bowl" and "knop" both collide with vessels already spoken for
calamus	branch
hastile	shaft	0269A: praedicatores de hastili procedentes — the branches come out of the shaft, so the two words must stay distinct
lilium	lily
candelabrum / lucerna	candlestick / lamp	Ex 25:37 sets the seven lucernae upon the candelabrum; collapsing either into "lampstand"/"light" loses the relation
⭐ emunctorium	snuffers	0270A: Emunctorium, et vasa, ubi quae emuncta* sunt exstinguuntur* — the noun and the participle must share a root, so "snuffers"/"snuffed off"
cortina	curtain
⭐ sagum	tent-cloth	the hard one. Douay renders both cortina and sagum "curtain", which erases a distinction CAPUT XXVI spends two columns on (ten cortinae of fine linen under eleven saga of goats' hair). Sagum had to be a distinct English noun that is not covering (operimentum), veil (velum) or hanging (tentorium), all four of which stand within six lines of one another at 0272C.
cilicium / cilicinus	haircloth	0271C reads Cilicium poenitentium est; the saga cilicina are then "the tent-cloths of haircloth"
velum	veil
operimentum	covering
tentorium	hanging	Ex 26:36, the screen at the entrance
tabula (of the tabernacle)	board
tabula (of the Law; of the propitiatory)	tablet; board	0265C's tabulis lapideis are tablets; 0267C's tabula aurea is a board, because it is the same measure as the ark and belongs to the structure. No gloss connects the two senses.
⭐ basis	base	0274B is an etymology: Basis est fundamentum a Graeco βασεία, id est fundo, and the argument continues quasi bases a fundamento surrexerunt. "Base"/"foundation" keeps it; "socket" (Douay) breaks it.
ansa / ansula	loop
fibula	buckle	0272B: fibulae sunt aeneae, de vocali… metallo
columna / caput	pillar / head	0275C ties capita aurea to mentem qua reguntur; "capital" would hide it
byssus	fine linen
⭐ hyacinthus / ianthinus	hyacinth / violet	0272D glosses Hyacinthus coelestis est coloris, and 0271C has nec hyacinthus caeruleo colore decoratur — the colour is argued about, so the stone-name is kept; ianthinus, a different word for a neighbouring colour, gets a different English
coccus bis tinctus / purpura	scarlet twice dyed / purple
⭐ lignum setim	setim wood	0273A says outright nec translatum est nomen Hebraeum, quia vix apud alios invenitur hoc lignum — the gloss's point is that the name was not translated, so translating it would contradict the sentence it stands in
incastratura / incastrare	mortise
compago	framing	0274D: Recederet enim ab invicem compago tabularum

Two further conventions carried from the peers, unchanged: usque ad → "as far as" (italic preserved), and Migne's italic parenthetical citations ((BED., lib. I de Tabernac.…), [ID., quaest. 98.]) are carried verbatim and untranslated on the same footing as [n: …].

§7 · Readings examined and DECLINED, with the reason

0272C — nec ad altera declinabant, [cj:] DECLINED. The same sentence prints ad latera twice within four lines (ut ad latera descenderent; saga autem tectum et latera operiebant), and latera is plainly what the argument wants: the skins covered the roof only and did not come down to the sides. But altera is a real word occupying its slot, and the English it yields — "nor did they slope down to the other parts" — is vague, not false. Pattern 18a's bar is that the printed word must make the sentence say something the author is arguing against, and the 2026-08-18 census explicitly declined the mood/number/word-slip class this belongs to. Both copies print altera, so it is Migne's; conjecture recorded here, nothing in the text.
0263A — Solemnitatem vero mensis primitivorum, [cj:]/[var:] DECLINED. The Vulgate (Ex 23:16) has solemnitatem messis primitivorum, "the feast of the harvest of the firstfruits". Both copies print mensis. Declined on two grounds: it stands in Rabanus's own prose and not inside a quoted lemma, so Pattern 14 does not reach it; and mensis primitivorum construes and is coherent — the sentence three lines above has Pascha… celebramus mense novorum, which is very likely what drew the compositor's eye, and which also makes "the month of firstfruits" a phrase this author could have written.
0270D — omnes in ordine numeros pleno ordine numeraveris, [d:] DECLINED. in ordine… pleno ordine reads as a dittography in our twin. Both copies print it exactly so. Pattern 11 wants a plate doubling and this is one, but it is not an eye-skip: in ordine and pleno ordine do different work ("number them all in order, by a full order"), there is no neighbouring phrase for either to have been copied from, and nothing is intrusive. Rendered as printed.
0264B — unde justo, judicio Dei expulsi sunt. The comma after justo is on the plate, on both copies. Pattern 8: never delete a mark Migne prints. Carried into the English ("whence by the just, judgment of God they were driven out"), which reads oddly and is meant to.
0272D — ne inter pressuras deficiant, muniunt and 0274D — alius negaret. Both end without a terminal stop, on both copies. Pattern 8: never supply. The English stops where the plate stops, in both places. (The archive copy appears to set a stop at 0274D; Gallica does not, and that is a tenth archive substitution rather than a reading.)
0268A — Faciesque illi labium.,. The period and the comma are both on the Gallica plate; the archive drops the period. Migne's pointing, carried into the English lemma as printed. A patch removing it would have been wrong, and the archive copy alone would have produced one.
0265A — Heldad et Meldad for the Vulgate's Eldad et Medad (Nm 11:26). Both copies print Migne's forms; proper-name orthography, Pattern 9. No marker.
0271C — ne interiora tenera clara qualibet adversitate turbentur. The asyndetic pair tenera clara is on both copies and in our twin. Rendered literally ("lest the inner things, tender and bright, be troubled"); no emendation attempted.
0264D — Non ejiciam eos a facie and 0263D — Praecedetque te angelus. Both truncate the Vulgate's following words (a facie tua anno uno; angelus meus). Lemma truncation is this book's normal practice and is not a divergence.
0266C — a raised mark before nec nisi Domino donante laxanda. A small superscript-height mark stands there on the archive copy. It is not resolvable as either an ink speck or a foot-of-page key at the resolution available, and the foot of that page carries no note it could key to (checked zero, both layers, both columns). Recorded as unresolved. Nothing marked, nothing patched, no [cn:] proposed — a wrong key entered in Migne's own voice is worse than none.

§8 · Checked zeroes worth recording

Foot-of-page apparatus: CHECKED ZERO on all eight leaves, both columns, both layers (numbered Forte conjectures and the asterisk/letter cross-reference layer alike). This includes f138 / p143, where CAPUT XXVII begins mid-page — the position the chunking rule names as exactly where Migne sets notes mid-page. Nothing there either. The work's coverage stays "spot"; this is a fuller sample, not a "full" claim.
Guillemets: 0 « and 0 » in all five chunks, consistent with the lemma brief's statement that the whole work contains one opening guillemet, at 0192A, outside this range.
(cont.) on the CAPUT XXVI head of chunk 0034 is our chunk boundary and is left untranslated in the English head, per the runbook's known-false-positive 7.

Stint 0035–0039, 2026-09-08. CAPUT XXVII–XXXII: the court of the tabernacle, the priestly vestments, the consecration, the altar of incense, and the golden calf. Every column on which a marker stands was opened at the plate (Gallica bpt6k5505319w, leaves f138–f144, printed corner numbers read first on all seven, both columns of every leaf). Plate reads are recorded in data/briefs/8949-PLATE-READS-0035.json; structural divergences in data/briefs/8949-PENDING-TEI-PATCHES-0035.md.

Span count. The brief declares 266 marked spans for this range. My own Latin counts 266 italic spans, and 266 again counted paragraph-bounded, with zero guillemets in all five chunks — consistent with the brief's statement that Exodus's single « stands at 0192A, outside my range. Per-chunk: 56 / 81 / 23 / 64 / 42. The totals agree, and the paragraph-bounded recount found no unmatched delimiter anywhere, so they agree for the right reason.

Marker parity. Column anchors 8 / 16 / 4 / 10 / 9 and notes 30 / 47 / 16 / 31 / 19, each identical to the Latin twin in count and in order, and each matching the chunk's own noteCount. Paragraph counts 32 / 51 / 15 / 41 / 26, matching the Latin 1:1. Question marks exact against the twin in all five (0 / 0 / 0 / 1 / 3, marker content excluded). Zero em-dashes in all five chunks.

1. ⭐ THE HEADLINE — THE ARCHIVE COPY REPAIRED MIGNE THREE TIMES IN THIRTEEN COLUMNS

The 0005–0009 stint's finding was that four apparent defects of Migne's were ours — Corpus Corporum's digitization, not the plate — and it fired zero [sic:] on that ground. This range ran the same experiment and it came out backwards.

col	our twin	archive copy	THE PLATE	verdict
0277A	`Sicuttibi`	`Sicut tibi`	`Sicuttibi`	Migne's. `[sic:]` fired
0283C	`snpra`	`supra`	`snpra`	Migne's. Carried, no marker (§4)
0284B	`uque ad`	`usque ad`	`uque ad`	Migne's. `[sic:]` fired

Three times, our twin was faithful and the archive was the corrupt witness — and corrupt in the direction that looks like a correction. That is precisely the jbig2 symbol-substitution mechanism PL113-WITNESS-ENCODING.md describes, running in the repair direction: a 1-bit symbol-substituting codec replaces a rare, broken shape with the frequent well-formed word its symbol dictionary already holds. 8957 @0825A recorded one such repair; this is three in one stint.

⛔ The operational consequence, and it is the one thing from this stint worth carrying forward. I consulted the archive's OCR text layer across the whole range before fetching a single Gallica leaf, because the common brief calls it "nearly free." It is free, and on this range it was wrong at every letter-level site that mattered — and it also manufactured two false alarms of its own (Vers. 25 for VERS. 35, lib. X Moral. for lib. I Moral., both at 0285C) that cost a plate check each to refute. Had I treated its agreement as corroboration I would have withdrawn three genuine defects of Migne's as digitization noise; had I treated its disagreement as evidence I would have filed two false patches against a faithful twin. On this book it is a search index, not a witness. Both directions of the twin-vs-plate error are now attested in this work, one stint apart, which settles the general rule negatively: neither our twin nor the archive may be trusted at a letter without the photograph.

2. `[sic: …]` — two markers, both plate-confirmed, both non-words

2a. 0277A — `Sicuttibi` for Sicut tibi

Migne prints VERS. 8. — Sicuttibi in monte. (BEDA, ibid.). The fusion is unambiguous on the plate at 4× — normal letter spacing throughout the fused run, clear word space either side — and the Clementine reads sicut tibi in monte monstratum est (Ex 27:8), so this is broken type, not a variant reading. Pattern 10 governs: the run is carried whole.

Rendered As to thee [sic: Sicuttibi] in the mount. The marker takes no second italic marker, because it sits inside an already-italic lemma and nested italics are not expressible (Pattern 10). The English renders the sense the fused form is defective for, per the 2026-09-05 amendment to Pattern 7 — a lemma fragment, so no verb, negation or conjunction is quarantined and 7a⁗ has nothing to catch.

2b. 0284B — `uque ad` for usque ad

Migne prints (ID.) Possumus haec altaria interpretari, etc., | uque ad potest accipi. with the marginal band letter B standing beside the line. The s of usque is simply absent; the rest of the word is clean and the surrounding spacing normal.

Rendered as far as [sic: uque ad] it can be taken. — the sense first, Migne's type inside the marker, again per the amendment. ⚑ This one is worth noting as a class: usque ad is the abridgment formula that runs 102 times through my five chunks alone, so the defective instance is invisible by eye in a page of identical formulas and is found only by reading the plate against the twin word for word. A frequency argument would have buried it, not found it.

3. `[var: …]` — eight markers, ranked by what the gloss rests on

Every one stands on a column I read myself. Each is cited from sources/vulgate/clementine-flat.txt by book chapter:verse, not from recollection.

⭐ 3a. 0286A — Migne's citation names Psalm LXII for words that are Psalm XXVI, and HIS OWN NEXT CITATION PROVES IT

The best finding of the range, because the control is on the same column. Migne prints:

VERS. 2.-- Beseleel, etc., qui interpretatur in umbra Dei. q. d. Dominus illuminatio mea et salus mea quem (Psal. LXII).

The words are Ps 26:1 — Dominus illuminatio mea et salus mea : quem timebo? — and the lemma breaks off at quem, exactly where the citation intrudes. Eleven lines below, at 0286B, Migne quotes the second half of that same verse — Dominus protector vitae meae; a quo trepidabo — and cites it correctly as (Psal. XXVI).

So one verse is cited twice on one column, once right and once wrong, and the reader who trusts the printed reference is sent to a psalm that does not contain the words. This is the ✓-against-the-wrong-address class the 0015–0019 stint opened (0227D Amos IX for Jdt 16:3, 0230B Isa. LXV for Rom 10:21), and it is the strongest instance yet because no external witness is needed to see it: Migne refutes himself in the same paragraph.

⚑ Plate-confirmed, and it had to be. Our twin reads LXII; the archive's OCR reads liii. Neither is XXVI, so the citation was wrong on any reading — but which wrong number Migne printed is a fact about the plate, and the plate settles it: LXII, in small caps, unambiguous at 5×. Marker fired on the plate's reading.

⭐ 3b. 0283A — 1 Tim 3:6 without its ne, and the purpose clause loses its conjunction

Migne prints Non neophytum in superbiam elatus, in judicium incidat diaboli (I Tim. III) against the Clementine's Non neophytum : ne in superbiam elatus, in judicium incidat diaboli. The dropped word is not a negation — it is the conjunction that governs the mood of everything after it, 7a⁗ question 3 exactly. Without it, incidat is a subjunctive with nothing to subordinate it, and the sentence as printed reads as an assertion that the neophyte, lifted up in pride, does fall into the devil's judgment — the opposite of Paul's warning, and the opposite of what Bede is arguing (that the newly baptized must not be ordained at once).

Rendered as printed — "Not a neophyte, lifted up into pride, he fall into the judgment of the devil" — which does not construe smoothly in English either, and that is correct: the marker carries the Clementine's reading and the gloss's argument survives beside it. No repair was made inside the translation.

⭐ 3c. 0283A — the words are John 14, the citation says John IV

Migne prints Ego in patre et pater in me est: et qui videt me, videt et patrem (Joan. IV) — a conflation of Jn 14:10–11 (ego in Patre, et Pater in me est) with Jn 14:9 (qui videt me, videt et Patrem), cited as the fourth chapter. Plate-confirmed, citation included. Same class as 3a; ranked below it only because Migne supplies no internal control here.

3d. 0288B — Rom 11:33 inscrutabilia for incomprehensibilia, with an added enim

Migne prints Quam inscrutabilia enim sunt judicia ejus: against the Clementine's quam incomprehensibilia sunt judicia ejus. A lexeme substitution in the verse that closes the whole chapter of the golden calf, adduced by Augustine as the answer to why God punished a people Aaron had led astray. Both the substituted word and the intrusive enim are plate-confirmed.

3e. 0287A — Ex 32:6 comedere for manducare

The verse lemma of the gloss that follows it (Esus potusque ad lusum impulit). Lexeme substitution in a verse lemma, plate-confirmed.

3f. 0287A — Sirach 19:1 in the present where the Clementine has the future

Migne prints Qui modica spernit, paulatim decidit against qui spernit modica paulatim decidet. The maxim is adduced to prove that small negligences grow into great ones, and the gloss's own next sentence is a present-tense generalization (Si enim curare parva negligimus… etiam majora audenter perpetramus), so Migne's present is doing work — but it is not the received text, and the tense is the whole force of a sentence about consequences. Marker records the future. Word order also reversed in the first two words.

3g. 0287D–0288A — Ex 32:25 in ignominiam for propter ignominiam

Migne prints Spoliaverat enim eum Aaron in ignominiam sordis et inter hostes, etc. against spoliaverat enim eum Aaron propter ignominiam sordis, et inter hostes nudum constituerat. The preposition changes the relation from cause (by reason of the ignominy) to end (unto the ignominy) — which is not nothing in a gloss whose entire subject is how much of the guilt is Aaron's. Plate-confirmed across the column break; see §6 on the break itself.

3h. 0286D — Mt 22:40 tota lex for universa lex

Migne prints in quibus scilicet tota lex pendet et prophetae, woven into Augustine's sentence. Lexeme substitution; the least consequential of the eight, and marked because it is a verse quoted as a proof-text and the collation is the deliverable.

4. ⚠ `snpra` at 0283C — Migne's defect, carried, and DELIBERATELY not marked

The judgment call of this stint, and the merge should feel free to overturn it.

Migne prints (BEDA, ubi snpra, cap. 11.) at 0283B–C. Plate-confirmed at 5×: the n is unmistakable, and the u of ubi on the same line gives the control. It is a genuine non-word and genuinely his — the third of the three sites where the archive silently repairs him.

Why no [sic:]. The run sits inside one of Migne's own bibliographic citations, and our TEI happens to wrap that citation in <hi> rather than <note> (see the patches file, §"STRUCTURAL"), so it reaches the English as body italics. Following the peer's 0262A precedent I carried the whole citation verbatim and untranslated, with the column anchor inside it, exactly as an [n: …] would be carried. And [n: …] contents are never marked: they are Migne's apparatus, and Pattern 12 was written for type carried through under Pattern 7 or Pattern 10 — that is, for the author's text, not for a locator.

Why it is arguable. The word appears on the English page and a reader cannot tell it from our own error, which is the exact failure Pattern 12 exists to close. The counter-argument is that uque ad in §2b got a marker eleven columns later, and the only principled difference is that usque ad is body text the English translates while ubi supra is apparatus the English does not. I think that difference is real and load-bearing, but it is thin enough that I am flagging it rather than burying it.

▶ If the merge promotes those three citations to <note> (patches file), this question closes by itself — the run becomes [n:] content and is unmarkable by rule.

5. Candidates raised and REJECTED, with the reason

A report that lists only findings cannot be told from a report by a stint that never looked.

0279A justitae — raised from our twin as a non-word (justitiae), withdrawn at the plate: Migne sets the æ ligature, justitæ. The divergence existed only in our transliteration of a ligature, which is the maechaberis class of ADJUDICATIONS §3 exactly. Pattern 9. No marker, no patch.
0276D Craticulumque for the Clementine's Craticulamque (Ex 27:4) — plate-confirmed, and a real divergence, but the difference is grammatical gender and English has no exponent for it: "gridiron" is the same word either way, and Migne himself prints the feminine craticulam / craticula at 0283D, seven columns later, in the same book. Pattern 9's gender clause governs; a [var:] whose content read "Vulg. Craticulamque, feminine" would be apparatus noise on a point no reader can see. Declined, recorded.
0279B catenulas auri purissimi for catenulas ex auro purissimo (Ex 28:14) — plate-confirmed. A genitive for a prepositional phrase; the English is "little chains of purest gold" on either reading and no gloss rests on it. Declined, recorded.
0282A–B tunicas parabis lineas for tunicas lineas parabis (Ex 28:40) — plate-confirmed, and it recurs three times in the inventory (0282B twice, again at 0282B in chunk 0037), so it is Migne's settled habit rather than a slip. Transposition only. Declined, recorded.
0278D Ipsa quoque textura … ac purpura — the brief flagged this span as not found in the Clementine. It is found: Ex 28:8 reads Ipsa quoque textura, word for word, and the sole difference is ac purpura for et purpura. The mechanical search failed on the conjunction. Declined; flag explained rather than left open.
0283B Superexaltat misericordia judicium without the Clementine's autem (Jas 2:13) — plate-confirmed. A dropped connective inside a proof-text quoted as a tag. Declined.
0277C Columnasque habebit for columnas habebit (Ex 27:16) — an added enclitic. Declined.
0286C exisse for exiisse (Lc 8:46) — attested syncopation, Pattern 9. Declined.
0285B Onychen for onycha (Ex 30:34) — a Greek accusative in a different declension pattern, not a different word. Declined.
0276C VLRS. 3. — a worn small-cap E on the plate in a structural label. Section labels normalize in English (Pattern 9) and the Latin column preserves what our twin has. Declined, no patch.
0276B VERS 1. without the period after VERS — Migne's, on two witnesses (plate and archive agree). The 0201C rule. Never patch, never mark.
0286B VERS. VI. — Migne's own Roman numeral in small caps, the only one in the range, and our twin's capitals are the right rendering of small caps. Declined.
0285C VERS. 35. and (GREG., lib. I Moral., …) — two false alarms raised by the archive's OCR (Vers. 25, lib. X) and refuted at the plate; our twin is right in both. Recorded because they are the cost side of §1's ledger.

6. Cruces that changed nothing on the page but are worth the record

0285C videant. Migne prints quamvis quatuor elementa haec quatuor videant significare — an active where the sense plainly wants the passive videantur ("although these four seem to signify the four elements"), which is how every reader will take it and how the following clause-by-clause mapping of spice to element requires it to be taken. videant is a real Latin form in a real grammatical slot, so Pattern 12 has nothing to wrap and Pattern 18's test is not met either — the faithful English is hard, not misleading. Rendered actively as printed ("although these four may see the four elements to signify"), conjecture here. ⚑ My first draft read "seem to signify," which is a silent repair of exactly the kind 7a names, and I caught it only on the reread against the Latin — recorded because the rendering that produced it is the one that feels most obviously correct.
0286C de quo dicitur. A full stop where a colon is wanted, immediately before the quotation Et virtus Altissimi obumbrabit tibi. Plate-confirmed, and the archive prints it the same way. Preserved as printed (Pattern 8, never delete).
0284C imponitur and 0285C incensum sit both end their paragraphs with no terminal full stop. Migne's, plate-confirmed. Pattern 8 forbids supplying one; the English stops where he stops.
0282A Proh dolor! autem — the exclamation mark falls before the postpositive autem. Preserved; the English carries the mark ("But, alas for the grief!") without re-punctuating.
0287B Hi sunt dii tui in Augustine's gloss against the lemma's Isti sunt dii tui three lines above (Ex 32:8 has Isti). Both render "These are" and English has no exponent for the distinction. No marker; recorded so a later sweep does not read the English as a conformation.
0287D–0288A. The plate breaks igno-|miniam across the column, and our twin's [0288A] anchor sits inside the word. That is the faithful encoding of a column break, not a Pattern 10 split run and not a [d:] — the 8956 hyphen class, checked and cleared at the plate.
0288B VERS. 31. — Migne's verse number for Aut dimitte eis is correct; the Clementine also numbers aut dimitte eis hanc noxam as Ex 32:31. An off-by-one I expected to find and did not.

7. Conventions this stint fixed, and why

Inherited from cruces-0025.md and followed without exception: praeceptum = "precept", mandatum = "commandment" (ADJUDICATIONS §1; my range uses praeceptum only, at 0278A and 0286D, and both are rendered "precept"); doctores = "teachers" and rector = "ruler", matching chunks 0013–0028; Pattern 17 throughout.

⭐ cruces-0030.md did not exist for most of this stint, and landed just before I closed. My launch brief told me to follow its list; for the whole translation I could not, so every term below was fixed independently. I reconciled against its §6 table before closing, and the agreement is close to total — which is itself a useful datum, since two agents who could not see each other converged on tabernaculum = tabernacle, basis = base (both of us for the fundamentum etymology), byssus = fine linen, hyacinthus = hyacinth, coccus bis tinctus = scarlet twice dyed, purpura = purple, lignum setim = setim wood, vectis = bar, lamina = plate, corona = crown, tentorium = hanging, velum = veil, and ⭐ caput = head, which we each chose over the idiomatic "capital" for the same reason: the gloss reads the pillar-head as the mind that governs (0275C there, 0277B here).

⚑ One real conflict, found and fixed. Its table fixes tabula of the Law = "tablet" (reserving "board" for the tabernacle's tabulae). I had used "tables" at four sites in 0039 (0286D ×3, 0287C ×1). Swept to "tablet" before closing. Had cruces-0030 not landed in time, this range would have shipped the Decalogue's tabulae under a different English from CAPUT XXV's — the 11551 split exactly, one chapter apart.

⚑ One apparent conflict that is not one. Its table fixes calamus = "branch" (of the candlestick, Ex 25:31ff). At 0285A my calami is Ex 30:23's spice — Calami. Cinnamomi est virgultum — a different object under the same noun, rendered "cane". Two senses, two Englishes; do not sweep one onto the other.

⭐ superhumerale = "superhumeral", NOT "ephod." The launch brief's rule decided this: prefer the rendering that keeps a gloss's own argument intact. Bede's exposition at 0278C opens Quia in humeris onera portamus — "because on the shoulders we bear burdens" — and the whole gloss is an etymology of super-humerale from humerus. "Ephod" is the idiomatic English and the Douay's word, and it shares no root with "shoulder," so under it the gloss explains nothing. This is the 11551 stylus/epistylium case exactly. "Superhumeral" is attested English in vestment literature, and it keeps the argument visible.
rationale = "rational" (Douay's term, and the corpus already uses "rational" for ratio in chunks 0002, 0007, 0011, 0026, so the tie to reason is live in this book's English). "Breastplate" was considered and rejected for the same reason as "ephod": Bede at 0279C glosses it by what the priest must think (congruant studiose cogitare).
⭐ capita of pillars = "heads", not "capitals" (0277B). Bede's gloss turns on the pun: Capita non argentata, sed tota argentea … Sicut enim membra sine capite vivere nequeunt. "Capitals" is the correct architectural English and it silently kills the argument, since "capitals" and "head" share no visible root for a reader. Douay uses "heads" at Ex 38:17. ⚑ My first draft had "capitals"; caught on the reread.
hyacinthus = "hyacinth" throughout, for both the colour (0280B, 0281C) and the gem (0279B). Douay's "violet" is better English for the colour, but Migne uses one word for both and the 0281C gloss (spe coelestium praemiorum, quam hyacinthus significat) depends on the colour and the stone being the same thing.
thymiama = "incense", incensum = "incense" — Migne equates them himself at 0284A (Incensum vel thymiama), and that one site is rendered "Incense, or thymiama," keeping the Greek-derived word in italic only there, because that clause exists to equate two words and collapsing it would leave "Incense, or incense." Elsewhere both are "incense" and altare thymiamatis / altare incensi are both "the altar of incense," which is what they are.
balteus = "belt", cingulum = "girdle" — they stand five lines apart at 0282B–C and need to stay distinct.
vitta = "fillet" (0279B, 0281C); ora = "border"; capitium = "head-opening" (0280C), chosen over Douay's "hole for the head" because the gloss immediately reads the capitium as the neck (Talis ora collum sacerdotis per gyrum vestit) and "head-opening" keeps that anatomy.
tiara / cidaris / mitra = "tiara" / "cidaris" / "mitre", all three kept distinct at 0282A, where the gloss exists to equate them (Tiara quae cidaris et mitra vocatur).
tentoria = "hangings" (Douay); columnae = "pillars"; bases = "bases" rather than Douay's "sockets", because Bede closes 0278A on timore Dei velut aenea basi stabilierit — a base to stand on, not a socket to sit in.
opus polymitum = "embroidered work", polymitarius = "embroiderer" — paired at 0286A, where David's title is derived from Beseleel's craft (polymitarii nomen, quia Beseleel opere polymito tabernaculum construxit).
receptacula ignium = "receptacles of fire" (0276C), not the technical "firepans," because the gloss's turn is that the teachers themselves receptacula ignium fiunt.
usque ad = "as far as", in italic, following chunks 0025–0029 exactly. It occurs 102 times in my five chunks and is never varied.

Chunks 0040, 0041, 0042, 0043 — CAPUT XXXIII to the end of the book: the vision of God's glory, the renewed tablets, the shining face of Moses, the completion of the tabernacle, and the closing recapitulation that names Exodus in Greek, Hebrew and Latin. The last stint on this work.

Plate range Gallica bpt6k5505319w f144–f148, all five leaves opened, printed corner numbers read first on every one (287/288, 289/290, 291/292, 293/294, 295/296 — the f = ceil(column/2) map held with no quire drift). Apparatus companions: data/briefs/8949-PLATE-READS-0040.json, data/briefs/8949-PENDING-TEI-PATCHES-0040.md.

⚠ cruces-0030.md and cruces-0035.md were not on disk when I began — chunks 0030–0039 had not landed — so I built the tabernacle vocabulary from the shipped English of chunks 0000–0029 first, and reconciled against the two peer files when they appeared mid-stint. The reconciliation is clean: hyacinth (never "violet" for hyacinthus), scarlet twice dyed, propitiatory, candlestick / lamp, laver, snuffers, bar for vectis, ring for circulus, curtain for cortina, board for the tabernacle's tabula, setim wood, brass for aes, ark of the testament, Beseleel, prolepsis, precept for praeceptum, as far as for usque ad — every one of these is what I had already written.

Two departures, both deliberate and both flagged rather than taken silently:

sethim, not setim, at 0295A. cruces-0030.md's reason for keeping the word untranslated is that the gloss at 0273A says outright nec translatum est nomen Hebraeum. Migne spells it sethim here (plate-confirmed, f148, first text line), and the same reasoning that keeps the Hebrew name keeps the spelling Migne gives it. Not a divergence from the convention; an application of its own argument.
⚑ tabulae of the Law: I render "tables", and so does the stint that wrote the rule. cruces-0030.md's vocabulary table rules tabula (of the Law) = "tablet". But the shipped English of chunks 0000–0029 reads "tables of stone" at four sites, and chunks 0030–0039 — that stint's own chunks — read "tables of stone" and "the tables" at three more. So the rule as written was never applied, including by its author, and the house usage is tables for the Law and boards for the tabernacle. I followed the usage, which keeps my three sites (0291B ×2, 0291C) consistent with all seven shipped ones and with 0290C's duas tabulas lapideas. ▶ Owed at merge: ratify the usage and correct the table, or sweep all ten sites. Flagging it because a later sweep reading the rule and not the corpus would churn seven correct sites — and because ADJUDICATIONS §1 exists for exactly this shape of split.

§1 · Span count against the brief — 120 against 120, EXACT

The brief declares 120 marked spans, of which 0 open-ended, for chunks 0040–0043.

chunk	italic spans, regex	italic spans, paragraph-bounded	anchors	`[n:]`
0040	42	42	9	20
0041	69	69	12	41
0042	0	0	0	0
0043	9	9	11	11
total	120	120	32	72

The two counts agree, and I checked why rather than inferring it from the totals: no paragraph in any of the four files contains an odd number of asterisks, so neither an unmatched * running forward nor an unmatched one shifting the pairing backward can be cancelling an error against another (the 8950 0004–0007 false reconciliation is exactly that shape). Zero guillemets in all four chunks, consistent with the brief's statement that the work's single « stands at 0192A, far outside my range.

Anchors and notes are 1:1 with the Latin twin, in the same order, and each chunk's [n:] count equals its own frontmatter noteCount.

⚑ The English files carry more asterisk pairs than the Latin, by 6 in 0040, 9 in 0041 and 1 in

Every one of those is inside a [var: …] marker, where Migne's word or the Clementine's is quoted in italic; the body counts are 42 / 69 / 0 / 9, exact.

§2 · Marker tally

marker	count	columns
`[var:]`	8	0289C, 0290C ×2, 0291A, 0291B, 0292A, 0292C, 0294B
`[sic:]`	2	0290C, 0291C
`[d:]`	0	— (one drafted at 0289B and withdrawn at the plate)
`[ed:]`	0	—
`[cj:]`	0	—

Every marked column was opened at the plate by me. No marker in this stint stands on an unread column, and none stands on the archive copy alone.

§3 · ⭐ THE HEADLINE — this book's FIRST two `[sic:]`, and what made them safe to fire

Six stints have now run on Exodus and the running total of [sic:] was zero. Stint 0005–0009 put the reason on record: four readings that looked like Migne's broken type were all Corpus Corporum's, and its finding was that [sic:] is a public accusation against Migne that this book had not yet earned. I fired two. Both turn on a single letter, and both were settled the same way.

0290C — `iu` for in

Migne prints, in italic, inside the verse lemma of Ex 34:7:

VERS. 7. — Qui reddis iniquitatem patrum filiis ac nepotibus iu tertiam et quartam, etc.

Gallica f145 (photograph, corners read first, right column, 2.6×): iu, plainly, the second stroke carrying no shoulder.
Our Corpus Corporum twin: iu.
The archive copy: in iertiam — it reads the preposition as in and breaks tertiam instead.

That third reading is the whole argument. A 1-bit symbol-substitution codec producing a frequent, expected word (in) in place of a rare non-word, while manufacturing a defect one word along, is the documented mechanism of PL113-WITNESS-ENCODING.md — it is not evidence against Gallica, it is the behaviour ADJUDICATIONS §5 says to expect. Here Gallica and our twin agree, and it is the archive that needs explaining. Rendered under the 2026-09-05 amendment: the sense is carried and Migne's type sits inside the marker, unto [sic: iu] the third and fourth.

⚑ 7a⁗ check. The marker swallows the preposition. The English says unto in plain words beside it, so the clause keeps the direction the type was carrying.

0291C — `Isael` for Israel

VERS. 34. — Et tunc loquebatur ad filios Isael omnia quae sibi fuerant, etc.

Same shape, and cleaner: Gallica prints Isael in italic with no under-inking at that letter (the surrounding line is well inked — loquebatur, omnia, fuerant are all crisp at 2.6×), our twin prints Isael, and the archive prints Israel. The archive is again the copy moving toward the expected word, and again it is the one that needs explaining. to the children of Israel [sic: Isael].

⭐ The rule I applied, stated so it can be checked: a [sic:] fires in this book only where Gallica and our twin agree against the archive. Where our twin stands alone — which is where the other three candidates in my range fell — the finding is a TEI patch and the English renders the plate.

§4 · ⭐ THREE `[sic:]` CANDIDATES WITHDRAWN AT THE PLATE, and one `[d:]`

All four are filed in 8949-PENDING-TEI-PATCHES-0040.md. Listing them here because the class is the finding, not the individual sites.

col	our twin	the plate	what was withdrawn
0289A/B	`aliquam [0289B] quam partem`	`quod ali-` / `quam partem` — one word, hyphenated	a `[d: ]` dittography
0291B	`decem quater dueta`	`decem quater ducta`	a `[sic: dueta]`
0295A	`annuns erant infixi`	under-inked `annulis` (archive: `annulis`)	a `[sic: annuns]`
0296C	`de plagis-Aegypti`	`de plagis` / `Aegypti` — line break, no hyphen	a `[sic:]` on a fused run

The 0289A/B site is the one worth reading twice. From our twin alone it is a textbook Pattern 11 dittography — a run set twice across a column anchor, exactly the shape 0224D/0225A produced a correct [sic:] for earlier in this book. It is not. Our transform doubled the fragment while inserting the band anchor at Migne's hyphenated line break. A dittography marker there would have been an accusation founded entirely on our own markup, and nothing downstream could have seen it.

⚑ Note the direction of the 0296C case: the hyphen in plagis-Aegypti is ours, and it fuses two words the plate keeps apart. That is the 7a⁗-b fused direction — a single well-formed-looking token that no mechanical test sees. It was found by reading the line, and only by that.

§5 · `annuns` at 0295A — what I can and cannot say

Read at 16×. The plate shows annu, then three strokes, then s: a short minim, then a taller stroke, then a short minim. That is not annuns (which needs two strokes) and it is not cleanly annulis either, because the middle stroke's ascender is partly lost to inking.

What is established: the plate does not print our twin's word, and the archive reads the word as annulis, and the sense requires annulis — the bars are fixed in the rings so the boards can be lifted from station to station. The patch is proposed as annulis and the English reads "in whose rings". The residual uncertainty is here rather than buried: if a later reader with a better copy finds a third reading, the patch is the thing to revisit, not the English's "rings".

I fired no [sic:], because what I demonstrated is a difference between our twin and both witnesses — not a defect Migne carries.

§6 · The eight `[var:]`, ranked by what the gloss is BUILT ON

The brief's rule is that the divergences worth most are the ones an argument stands on. Ranked:

⭐⭐ 1. 0291A — immolatorum meorum (Ex 34:25), and Augustine's question exists only because of it

Migne prints the Vulgate lemma and then Augustine's own text four words later:

VERS. 25. — Non immolabis, etc. (AUG., quaest. 162 et 163.) Quid est quod ait: Non occides super fermentum sanguinem immolatorum meorum? An illa hoc loco dicit immolata sua, quae per pascha occiduntur…

The Clementine (Ex 34:25) has Non immolabis super fermento sanguinem hostiae meae. Augustine's next sentence asks whether immolata sua — "his immolated things", a noun — means the paschal victims. The Vulgate's hostiae meae gives him no such noun, and the question cannot be put from it. Conform the lemma and the paragraph loses the word it is about. Plate-confirmed at the head of f146's left column, the first line under the running head.

⭐⭐ 2. 0291B — Videns autem Aaron and the vanished horned face (Ex 34:30)

VERS. 30. — Videns autem Aaron, etc. (ORIG., hom. 12 in Exod.) Scriptum est quia vidit Aaron et omnes filii Israel Moysen**…

The Clementine has Videntes autem Aaron et filii Israel cornutam Moysi faciem. Two divergences in one line, and they run together: Migne's participle is singular, agreeing with the gloss's vidit, and what they see is Moses himself, not the cornuta facies — Jerome's horned face, the most famous word in this verse. Origen's homily is about the face being glorified, and it continues in lege vultus Moysi glorificatur, licet velamine contegatur. The Vulgate's horns are not what he is expounding. One marker on the lemma; the gloss's Moysen is named inside it rather than marked twice.

⭐ 3. 0289C — prae omnibus (Ex 33:17), which the whole question hangs from

VERS. 17. — Et teipsum novi, etc. (AUG., quaest. 153.) Quoniam scio te prae omnibus, etc. Nunquid Deus plus aliqua scit et aliqua minus?

The Clementine has et teipsum novi ex nomine. Augustine's question — does God know some things more and some less? — is generated by the comparative force of prae omnibus, "above all". Ex nomine has no comparison in it at all, and from the Vulgate the question is unmotivated. Plate-confirmed at 2.6×.

4. 0292A — Rom 13:14, in deliciis for in desideriis

Induite vos Dominum Jesum Christum, et carnis curam ne feceritis in deliciis (Rom. XIII).

Clementine: sed induamini Dominum Jesum Christum, et carnis curam ne feceritis in desideriis. Two changes, and the second is lexical, not morphological: deliciis (delights, pleasures) for desideriis (desires). Origen adduces the verse for nobiliora indumenta — nobler garments — so the imperative Induite vos is doing work here that the passive induamini does less plainly. Plate-confirmed; the archive corroborates in deliciis.

5. 0294B — the ✓-against-the-wrong-ADDRESS class, live again

unde: Caput viri Christus est (II Cor. XI).

The words are 1 Cor 11:3's — and reordered: the Clementine has omnis viri caput, Christus est. Migne's own citation names Second Corinthians. This is the class the brief names from 0227D and 0230B, and it is the reason a ✓ in the pre-scan is the beginning of a check and not the end: the inventory scored this span ⚠ and the address it should have matched was one book away from the one Migne prints. Plate-confirmed at f147, right column; the archive corroborates the II Cor. attribution, so it is not our transcription.

6–7. 0290C — Augustine's Exodus text, twice in one paragraph

Et protegam (AUG., ubi supra.) et tegam manu mea super te, etc., usque ad deinde contexit quid Dominus postea dicat: Excide tibi duas tabulas lapideas, etc.

Clementine Ex 33:22 et protegam dextera mea, donec transeam; Ex 34:1 Praecide, ait, tibi duas tabulas lapideas. Migne sets the Vulgate's Et protegam and Augustine's tegam manu mea super te in one breath, which is itself the evidence. Both plate-confirmed on the same 2.6× crop. Ranked below the first five: the glosses they head are abridged away by usque ad, so nothing visible on the page is built on either reading.

8. 0292C — Ex 36:2, the Vulgate and Augustine side by side again

Et qui sponte, etc. (AUG., quaest. 170 in Exod.) Et omnes qui sponte vellent ire ad opera, ut consummarent ea, etc.

Clementine: et qui sponte sua obtulerant se ad faciendum opus. Plate-confirmed. Same class, same ranking reason.

§7 · The class stated once, and the divergences I did NOT mark

The class: chapters XXXIII–XXXVI are carried by Augustine's Quaestiones in Exodum and by Origen's homilies, and both quote a Latin Exodus that is not the Vulgate. That is a fact about the sources, not about Migne's press, and marking every instance would turn the page into apparatus. I marked five sites in the class and state it here for the rest.

Declined, each with its reason:

0290C Excide/Praecide — marked (6 above), but ranked low for exactly this reason.
0291A Non coques agnum in lacte matris suae vs the lemma's own haedum (Ex 34:26). Migne prints both, three words apart, the Vulgate's kid and Augustine's lamb. The gloss is then abridged away by usque ad, so nothing on the page turns on the swap. No marker; Migne's own juxtaposition is the disclosure.
0291C Alia littera: Sumite a vobis ipsis redemptionem Domino and 0291D Alia littera: Unusquisque vestrum — ⛔ Migne labels these himself. §11's rule holds: take no marker where he has already marked it.
0289B Lucem habitat inaccessibilem vs 1 Tim 6:16 lucem inhabitat inaccessibilem. Plate-confirmed as habitat. English has no exponent for the prefix — "dwelleth in light inaccessible" renders either — so a marker would announce a divergence the reader cannot see. Pattern 9's logic, no marker.
0290B Super hanc petram aedificabo Ecclesiam — Mt 16:18 has Ecclesiam meam. Migne abridges and closes with a full stop, as he abridges everywhere in this work. Not a divergence, an abridgment. No marker.
0290A Ecce, inquit, locus est apud me vs Ex 33:21 Ecce, inquit, est locus apud me. Word order only, no sense at stake. No marker.
0288C Audiens populus vs Ex 33:4 Audiensque populus. A lemma head clipped of its enclitic. No marker.
0293A Fecitque candelabrum vs Ex 37:17 Fecit et candelabrum ductile; 0292D Fecit autem Beseleel arcam vs Ex 37:1 Fecit autem Beseleel et arcam; 0292A Pilosque caprarum vs Ex 35:6 pilos caprarum. Three lemma heads differing by a connective. No markers.
0293C Locutusque est Dominus ad Moysen dicens: Mense primo — scored ⚠ only because it conflates Ex 40:1 and 40:2, which the pre-scan searches as one line. Verbatim against both. No marker.

§8 · Renderings fixed here, for the merge's conventions header

usque ad → as far as, italic, following every shipped chunk of this work.
praeceptum → precept throughout (ADJUDICATIONS §1's provisional convention, applied); mandatum keeps commandment. In my range the noun does not fall, but the verb praecipio does, three times — 0288D praecipiente Domino, 0290D Cum de Sabbato praeciperet, 0291A praecipit ne tunc sit in domo fermentum. I rendered these "the Lord enjoining it", "when he was giving precept concerning" and "gives precept that" — keeping "command" out of the verb as well, so the noun convention is not undercut eleven words away by its own cognate. ⚑ This is a decision the merge should ratify or overturn once, not per stint.
Nostis at 0291D and Unusquisque vestrum at 0291D are plural → "you", while inveniat aliquid tuum … sibi te defendit in the same sentence is singular → "thine", "thee". Pattern 17 exactly, and the shift is visible on the page as Origen turns from the congregation to the single hearer.
sethim (0295A), not setim. Chunks 0000–0029 render setim wood because Migne spells it setim there; here he spells it sethim and the English follows the plate.
hyacinthum → hyacinth, never "violet" — inherited from the shipped chunks, per the note at the head of this file.
paxillus → pin (0293B, Ex 38:20). New here; the word does not fall in any earlier range. Chosen over "stake" and "peg" because the gloss turns on the pins being in modum linguae facti — made after the manner of a tongue — and on their being what the cords are tied back to at the foot while little loops hold them above. "Pin" is also the Douay word at Ex 38:20, so it collides with nothing already fixed. ansula → little loop, kept distinct from circulus (ring), which stands eleven lines away at 0292D.

§9 · Cruces carrying no marker

0296A spiritualis exitus sanctus sanctorum de mundo. Plate-confirmed (f148, right column, 2.6×). sanctus is nominative singular standing beside sanctorum; the phrase will not construe smoothly, and the obvious conjecture is that sanctus is a compositor's echo of the genitive beside it. Rendered as printed — "and the spiritual going out, holy, of the saints from the world" — with no marker: sanctus is a real word occupying a real slot (an adjective in apposition to exitus), which is precisely what Pattern 12 says takes no [sic:]. The conjecture lives here, and the awkwardness on the page is Migne's.
0291C de servitute saeculi liberati libertatem scientiae requiramus. The archive reads liberari. I did not open that line on Gallica and take no position: liberati is sound, agrees with the subject of requiramus, and the archive may never be a sole witness for a letter. Recorded as an open, immaterial disagreement so a later reader does not think it was missed.
0292D [n: (AUG., quaest. 171)]. — the note's closing period stands outside the bracket, where every other note in the work closes inside. Migne's parenthesis is (AUG., quaest. 171.) on the plate. This is a transform artifact of the same family as §3 of the patches file, but it changes nothing a reader sees and no count; recorded, not patched.

§10 · The end of the book, and the leaf it ends on

f148 is the work-division leaf. Exodus ends at 0296C and LIBER LEVITICUS begins below it; the foot of that same leaf is already Leviticus matter (Tribus Levi tabernaculi ministerium sortita est; Latine Offertorius vel sacrificativus dicitur), which is an independent confirmation of where the division falls. Nothing past 0296C was translated.
The Hebrew at 0296A was not re-opened and carries no marker. ואלה שמות stands verbatim in the English beside Migne's own transliteration veelle semoth, which is what gets translated — mirroring the banner at 0183A. It is unitalicized in the English because our twin sets no italics there and span structure is 1:1.
CAPUT XXXIX (chunk 0042) is a row of dots on the plate too. Migne prints the chapter head and a single line of widely spaced points, with no gloss. Verified at f147, left column. It is his blank, not lost digitization, and it correctly takes no [ed: …].
The foot of all five leaves is a CHECKED ZERO, both columns, all three key shapes (number, asterisk, raised letter). Recorded as a result in the plate-reads file. Per ADJUDICATIONS §4 this says nothing whatever about any leaf I did not open.

Liber Exodus

§0 · WORK-WIDE CONVENTIONS — settled at merge, and binding on any later sweep

§1 · Span count against the brief — 94 against 94, EXACT

§2 · Marker tally

§3 · [cj:] — 0183A, corporalis where the argument requires spiritualis

§4 · Two [sic:] candidates DECLINED, with the evidence

§5 · The 7a″ collation — 8 [var:], and every rejection reasoned

Fired

Rejected, and why — the ⚠ entries that are innocent

§6 · Readings rendered literally, no marker (Pattern 7 / Pattern 9)

§7 · Recovered Hebrew and the book-title banner — carried, not touched

§8 · Structural verification run before shipping

§1 · Span count against the brief — 133 against 133, EXACT

§2 · Marker tally

§3 · The 7a″ collation — 6 [var:], and why each fired

Divergences examined and DECLINED, with the reason

§4 · Four [sic:] candidates WITHDRAWN at the plate — all four were OURS, not Migne's

§5 · Examined at the plate and deliberately NOT changed

§6 · One case where the plate corrected a label — Vers. 9. at 0202D

§7 · Readings rendered literally, no marker

§8 · Conventions aligned with the peer stints

§9 · Structural verification run before shipping

1. ⭐ THE HEADLINE FINDING — Rabanus's Decalogue is not quoted from Exodus 20

2. [var:] markers standing outside the Decalogue class

3. ⛔ CANDIDATES WITHDRAWN AT THE PLATE — three, and each would have been a false claim

4. Candidates examined and DECLINED (with the reason, so they are not re-raised)

5. Migne's mis-bound references — his, and therefore untouched

6. Renderings decided, and why

7. Two observations that need a witness we do not have

8. ⛔ TWO ITEMS FOR MERGE-TIME RECONCILIATION — they cross stint boundaries

8a. praeceptum is being rendered two ways inside one enumerated series

8b. The first precept diverges too, and it is in a peer's chunk

§1 · Span count against the brief — 143 against 143, EXACT

§2 · Marker tally

§3 · The one [sic:] — 0225A, ververba

§4 · The 7a″ collation — 33 [var:], and why each fired

§5 · Divergences examined and DECLINED, with the reason

§6 · Readings examined at the plate and left alone

§7 · Register notes

§1 · Span count against the brief — 186 against 186, EXACT

§2 · Marker tally

§3 · The 7a″ collation — ranked by whether the gloss is BUILT on the divergence

Tier 1 — the gloss expounds a word its own lemma does not print

Tier 2 — Migne's address names one place and his printed words are another's

Tier 3 — real divergences, sense affected, gloss not built on them

§4 · The two markers that are claims about Migne

§5 · Six candidate [sic:] withdrawn — every one was ours

§6 · Declined, and the reasons

§7 · Two things about this range worth carrying forward

1. ⭐ THE HEADLINE — the Decalogue's FIRST precept, and it is now marked

2. The rest of the [var:] markers — twelve in all, with the bar I applied

2a. The Decalogue and its lemmata

2b. Proof-texts whose argument stands on Migne's word

3. ⛔ ZERO [sic:], and that is the finding

4. Migne's own oddities — recorded, untouched, and none of them ours

5. ⚠ Two readings left open, and neither was decided on one witness

6. ⭐⭐ A FINDING ABOUT THE BOOK, NOT ABOUT MY RANGE — Migne has a foot-of-page note layer here

7. Renderings decided, and why

8. ⭐ FOR THE MERGE — new evidence on praeceptum / mandatum, and it cuts one way

9. Declined as a class, once, so it is not re-raised

§1 · Span count against the brief — 266 against 266, EXACT

§2 · Marker tally

§3 · ⭐⭐ THE HEBREW LAYER OF THIS BOOK IS NOT COMPLETE — three more words, one of them load-bearing

§4 · ⛔ ZERO [sic:] — and the reason is that the ARCHIVE COPY IS THE WORSE WITNESS ON THIS QUIRE

§5 · The [var:] list — ranked by how much of the gloss stands on it

⭐ 1. 0267D — Hab 3:2 · In medio duorum animalium innotesces

⭐ 2. 0270B — Ex 26:1 · Tabernaculum vero ita et fiet

⭐ 3. 0269B — Ct 6:10 · Descendit in hortum nucum

⭐ 4. 0265A — Nm 11:16 · Assume tibi septuaginta viros

5. 0273B — the citation names Matth. XXI and the words are Mt 26:56

6. 0268A — Ct 2:4 · Introduxit me rex in cellam vinariam, et ordinavit in me charitatem

7. 0268A — VERS. 23. printed twice, the second over Ex 25:24

8. 0269A — 2Cor 13:13 · sit semper cum omnibus vobis

9. 0262D — 2Cor 9:6 · in benedictionibus, in benedictionibus et metet

10. 0274D — Ex 26:24 · Eruntque sibi conjunctae

Checked and DECLINED — divergences that do not change the sense

§6 · ⚑ THE TABERNACLE VOCABULARY — fixed here, and chunks 0035–0039 should follow it

§7 · Readings examined and DECLINED, with the reason

§8 · Checked zeroes worth recording

1. ⭐ THE HEADLINE — THE ARCHIVE COPY REPAIRED MIGNE THREE TIMES IN THIRTEEN COLUMNS

§3 · `[cj:]` — 0183A, `corporalis` where the argument requires `spiritualis`

§4 · Two `[sic:]` candidates DECLINED, with the evidence

§5 · The 7a″ collation — 8 `[var:]`, and every rejection reasoned

§3 · The 7a″ collation — 6 `[var:]`, and why each fired

§4 · Four `[sic:]` candidates WITHDRAWN at the plate — all four were OURS, not Migne's

§6 · One case where the plate corrected a label — `Vers. 9.` at 0202D

2. `[var:]` markers standing outside the Decalogue class

§3 · The one `[sic:]` — 0225A, `ververba`

§4 · The 7a″ collation — 33 `[var:]`, and why each fired

§5 · Six candidate `[sic:]` withdrawn — every one was ours

2. The rest of the `[var:]` markers — twelve in all, with the bar I applied

3. ⛔ ZERO `[sic:]`, and that is the finding

§4 · ⛔ ZERO `[sic:]` — and the reason is that the ARCHIVE COPY IS THE WORSE WITNESS ON THIS QUIRE

§5 · The `[var:]` list — ranked by how much of the gloss stands on it

7. 0268A — `VERS. 23.` printed twice, the second over Ex 25:24

2. `[sic: …]` — two markers, both plate-confirmed, both non-words

2a. 0277A — `Sicuttibi` for Sicut tibi

2b. 0284B — `uque ad` for usque ad

3. `[var: …]` — eight markers, ranked by what the gloss rests on

4. ⚠ `snpra` at 0283C — Migne's defect, carried, and DELIBERATELY not marked

§3 · ⭐ THE HEADLINE — this book's FIRST two `[sic:]`, and what made them safe to fire

0290C — `iu` for in

0291C — `Isael` for Israel

§4 · ⭐ THREE `[sic:]` CANDIDATES WITHDRAWN AT THE PLATE, and one `[d:]`

§5 · `annuns` at 0295A — what I can and cannot say

§6 · The eight `[var:]`, ranked by what the gloss is BUILT ON